new

Get trending papers in your email inbox!

Subscribe

Daily Papers

byAK and the research community

Nov 20

TeamBench: Evaluating Agent Coordination under Enforced Role Separation

Agent systems often decompose a task across multiple roles, but these roles are typically specified by prompts rather than enforced by access controls. Without enforcement, a team pass rate can mask whether agents actually coordinated or whether one role effectively did another role's work. We present TeamBench, a benchmark with 851 task templates and 931 seeded instances for evaluating agent coordination under operating system-enforced role separation. TeamBench separates specification access, workspace editing, and final certification across Planner, Executor, and Verifier roles, so that no role can read the full requirements, modify the workspace, and certify the final answer. Prompt-only and sandbox-enforced teams reach statistically indistinguishable pass rates, but prompt-only runs produce 3.6 times more cases where the verifier attempts to edit the executor's code. Verifiers approve 49% of submissions that fail the deterministic grader, and removing the verifier improves mean partial score in the ablation. Team value is also conditional. Teams benefit when single agents struggle, but hurt when single agents already perform well. A 40-session human study under the same role separation shows that our benchmark exposes interaction patterns that pass rate misses. Solo participants work through the task directly, human participants paired with agents often collapse into quick approval, and human teams spend more effort coordinating missing information across roles.

  • 12 authors
·
May 7

SemaPLC: A Project-Grounded, Verification-Gated Agent Harness for PLC Code Generation

Programmable logic controllers (PLCs) run industrial plants, and large language models can already generate independent program organization units (POUs) for them. Whether such logic integrates into an existing PLC project and then runs correctly has been checked only in limited tests. We present SemaPLC, a project-grounded and verification-gated agent harness assembled from conventional tools but governed by a strict completion rule. Rather than stopping when the model judges its own output adequate, SemaPLC declares a task complete only when logged external checks confirm it. Those checks cover the specification, the compilation, and the behavior on a live runtime. On 117 independent-POU tasks matching existing benchmarks, it attains the highest strict verified pass rate on all seven models (72.6\% mean). On a project-context track of 65 tasks whose generated logic must compile and run inside a real project, it attains the highest mean on integrated compilation, static behavior, and dynamic behavior. Of the three layers, dynamic behavior is the most revealing. We measure it by deploying the generated and the reference logic to a live PLC runtime and comparing their executed traces. All methods fall within 10 static points of one another, whereas dynamic scores separate them sharply, from 22.4 to 31.4 for the baselines against 52.2 for SemaPLC. Overall, our verification-gated harness raises the mean at every layer and most sharply at runtime. Execution, not static scoring, is the faithful test of whether generated control logic actually works. SemaPLC is open-sourced at https://github.com/midea-ai/SemaPLC.

What Can a Leaderboard Certify? Compositional Controllability for Fair Evaluation and Training of Biomedical Literature-Review Agents

Leaderboards rank long-horizon agents by their final outputs. Yet a higher score alone does not establish whether two systems are comparable or which stage accounts for the difference. Unequal evidence, inputs, or budgets can affect scores, and statistical corrections do not remove this mismatch. We introduce compositional controllability to address these questions. A comparison window covers one stage, several stages, or the whole agent. Our central result bounds the gap between observed and controlled score differences using only nuisance outside the window. This yields an admissibility test applied before scores are inspected. Inadmissible comparisons are refused. For admissible pairs, an ordering is certified only when the score gap exceeds the combined sampling and nuisance radii; otherwise, it remains undecided. These decisions give each system a rank interval. We introduce BioLitBench, a benchmark of 2,042 biomedical articles represented as structured claim graphs. Among seven published pipelines, a conventional statistical analysis declares a winner in 14 of 21 pairwise comparisons. Yet the top-ranked system alone received the target review's bibliography. To isolate pipeline performance, our test requires matched inputs and a fixed backbone model. It refuses 11 of the 21 comparisons, including every comparison involving the top-ranked system. Seven of the 14 conventional conclusions fall within these refused pairs. The same comparison windows support stage-level training. We train SCRIBE on Qwen3.8-27B using rewards measured at each stage's exit. Under matched evidence, SCRIBE achieves a certified rank interval of [1,2], with certified advantages over all evaluated published pipelines and the evaluated Claude and OpenAI agents. Under same pool, SCRIBE matches the strongest published retriever and is certified above three published pipelines.

  • 7 authors
·
Sep 25

WANDR: A Benchmark for Wide and Deep Research

WANDR (Wide ANd Deep Research) is a benchmark of 500 realistic, challenging data-collection tasks for research agents. Each task requires a system to discover a large set of entities that satisfy specified criteria (breadth), investigate each entity through multiple coordinated web searches (depth), and return independently verifiable records with supporting sources and excerpts. Tasks are represented as qualification key hierarchies that specify the entities, relationships, evidence, and required count at each level; a hierarchy with n companies, m employees per company, and k sources per employee requires n x m x k records. This structure supports diverse workflows such as market mapping, due diligence, literature review, product comparison, and talent sourcing, with targets ranging from dozens to thousands of records. WANDR replaces static gold answer sets with task-specific judges that refetch cited pages and verify each record against its evidence, allowing evaluation of current and changing facts. Record verdicts are aggregated into soft and hard precision, recall, and F1 scores that distinguish factual quality, coverage, and hierarchical completeness. The tasks are derived from de-identified product-usage logs and produced through a semi-automated pipeline with automated checks, empirical audits, and human review where needed. We evaluate six production research systems and find that the benchmark is far from saturated: at high effort, the strongest system reaches only 0.363 soft F1 and 0.133 hard F1. Performance degrades as target volume and hierarchy depth increase, with incomplete discovery, missing enrichment, and incomplete evidence construction remaining major bottlenecks. The benchmark and evaluation harness are available at https://github.com/perplexityai/wandr.

  • 10 authors
·
Aug 13

Where Does Agent Reliability Come From? A Cross-Benchmark Decomposition of Verification Loops, Specialist Models, and Scaffolding in a Production Enterprise Agent

Multi-step enterprise agent tasks fail in a characteristic way: single-pass inference has no checkpoint between deciding an answer and committing to it. We study one production system (Leni) whose architecture installs such checkpoints: verification loops (execute, observe, compare, correct) staffed by lightweight task-specialized post-trained models. We evaluate the unmodified production configuration on three public benchmarks stressing distinct failure modes: SpreadsheetBench Verified (silent computation error), BullshitBench v2 (premise confabulation), and the GAIA validation split (cascade error over long tool chains). The full system improves over its frontier base model by +11.0 percentage points on SpreadsheetBench (91.25% vs 80.25%, n=400, p<0.001), +7 to +10 percentage points on BullshitBench (98% vs 91%, n=100), and roughly +15 points on GAIA validation (75.2% pass@1, n=165; 83.0% best-of-k). Our central contribution is a decomposition of that uplift: most of it comes from scaffolding, routing, and specialist models rather than from the verification step itself, whose isolated contribution is small (+1.5 points) but concentrated at the top of the score distribution, where it converts otherwise-failing tasks. We instrument the loop end-to-end, yielding an empirical verifier confusion matrix (catch rate about 0.20, fix rate 0.75, no false-alarm regressions) that grounds a compounding-reliability model. Specialist-swap ablations suggest that the loop's value depends on who observes it: replacing the small trained verifier with the generating frontier model eliminates most rescues. A valid-premise control shows zero over-rejections in 100 expert-level questions.

  • 1 authors
·
Jul 18

AlphaNIST: Emergent Assembly Skills from Binary Task Rewards

The NIST Assembly Task Board #1 remains a challenging benchmark for learning-based methods in robotic manipulation. Over short horizons, it demands precise, contact-rich control of small, tight-fitting parts. Over long horizons, it requires recovery and coordination across extended action sequences. Demonstrations, task decomposition, and dense progress rewards have enabled research advances by providing guidance on how to assemble. But how much of the strategy must be specified? The progression from AlphaGo to AlphaGo Zero showed that RL at scale can discover superhuman strategies without prescribing how to win. AlphaNIST asks whether a sparse binary task reward and task diversity admitted by this task-agnostic reward can similarly let a robot discover how to assemble. In simulation, we train a single policy from scratch across all 20 assembly tasks on the NIST Assembly Task Board #1, reaching 98.2% overall training success without demonstrations, progress shaping, scripted primitives, or per-task tuning. The framework also trains policies to complete ten assembly pairs in a single episode without a prescribed order, with task success rewarded only when all ten pairs are simultaneously complete. The learned policies discover dexterous and planning-like behaviors: choosing grasps for later assembly, using surrounding geometry to reorient parts, and preparing one part, assembling another, then returning to finish the first without an explicit planner. For videos, see our website: https://alpha-nist.github.io/

  • 15 authors
·
Oct 8

TenetBench: A Clause-Anchored Russian Corpus for Evaluating Language Models against Professional Codes of Ethics

Codes of professional ethics guide and restrict specialists, who increasingly work with assistants built on large language models. Such an assistant can breach the user's code, or help the user breach it, without the user noticing. Evaluations reporting broad categories of harm cannot show how often this happens or which provision is breached. TenetBench pairs 3,433 requests conflicting with clauses of fifteen Russian-language professional codes with benign twins that differ only in the legitimacy of what is asked. An automated pipeline generates the tasks from the text of a code, so the corpus grows with every code added and a provision yields new tasks on every run. It also produced tasks and human-validated verdicts from two English codes of other jurisdictions. In human validation of the frozen pipeline, specialists rated 0.995 of the conflicting tasks plausible, and the model judge agreed with the adjudicated label at Gwet's AC1 of at least 0.894 on 1,860 elements. We evaluated fourteen model and reasoning-mode combinations on whether they produce the conflicting text, decline or warn. Eleven of the fourteen produce conflicting content on most such requests, mostly without a warning, and nearly three warnings in four accompany conflicting content. The models that conform more do so by declining while still serving the legitimate twin, and enabling reasoning halved the refusals of two models. We hope the dataset and pipeline will help evaluate how language models observe professional norms and let diagnostic datasets be built for particular professions and organisations.

  • 5 authors
·
Oct 20 6

The Art of Building Verifiers for Computer Use Agents

Verifying the success of computer use agent (CUA) trajectories is a critical challenge: without reliable verification, neither evaluation nor training signal can be trusted. In this paper, we present lessons learned from building a best-in-class verifier for web tasks we call the Universal Verifier. We design the Universal Verifier around four key principles: 1) constructing rubrics with meaningful, non-overlapping criteria to reduce noise; 2) separating process and outcome rewards that yield complementary signals, capturing cases where an agent follows the right steps but gets blocked or succeeds through an unexpected path; 3) distinguishing between controllable and uncontrollable failures scored via a cascading-error-free strategy for finer-grained failure understanding; and 4) a divide-and-conquer context management scheme that attends to all screenshots in a trajectory, improving reliability on longer task horizons. We validate these findings on CUAVerifierBench, a new set of CUA trajectories with both process and outcome human labels, showing that our Universal Verifier agrees with humans as often as humans agree with each other. We report a reduction in false positive rates to near zero compared to baselines like WebVoyager (geq 45\%) and WebJudge (geq 22\%). We emphasize that these gains stem from the cumulative effect of the design choices above. We also find that an auto-research agent achieves 70\% of expert quality in 5\% of the time, but fails to discover all strategies required to replicate the Universal Verifier. We open-source our Universal Verifier system along with CUAVerifierBench; available at https://github.com/microsoft/fara.

  • 5 authors
·
Apr 4

How Complete Must an Environment Be to Train a Capable GUI Agent?

Graphical user interface (GUI) agents are widely trained in executable GUI environments. To lower cost, build environments in batches, and reduce manual effort, each work trades away part of the environment through its construction choices. We describe these choices along three dimensions: reward coverage, the goal-reaching trajectories the verifier rewards; data coverage, the application data the agent is exposed to beyond what the task reads or writes; and feature coverage, the application features beyond what the task uses. Existing works set each dimension on engineering grounds, but they do not measure its effect on training. How complete must an environment be to train a capable agent, and along which dimensions? In MobileGym, we adjust each dimension separately, train Qwen3-VL-4B-Instruct, and evaluate every policy in the original full environment. A verifier that rewards only one reference path lowers success by 10.53 points, and removing unneeded records also lowers success. Pruning unneeded features instead raises success from 51.61% to 56.13%, entirely on tasks whose environments it left unchanged. The gain reverses when each episode is weighted equally and vanishes when only episodes on the pruned tasks are balanced. Without pruning, length weighting there reinforces long detour successes that erode skills on other tasks. Following these findings, we build MobileGym-Lite, which keeps reward and data coverage complete and prunes unneeded features. It outperforms Full in the full environment (55.48% versus 51.61% success) with lower construction cost.

  • 8 authors
·
Nov 17

HLE-Verified: A Systematic Verification and Structured Revision of Humanity's Last Exam

Humanity's Last Exam (HLE) has become a widely used benchmark for evaluating frontier large language models on challenging, multi-domain questions. However, community-led analyses have raised concerns that HLE contains a non-trivial number of noisy items, which can bias evaluation results and distort cross-model comparisons. To address this challenge, we introduce HLE-Verified, a verified and revised version of HLE with a transparent verification protocol and fine-grained error taxonomy. Our construction follows a two-stage validation-and-repair workflow resulting in a certified benchmark. In Stage I, each item undergoes binary validation of the problem and final answer through domain-expert review and model-based cross-checks, yielding 641 verified items. In Stage II, flawed but fixable items are revised under strict constraints preserving the original evaluation intent, through dual independent expert repairs, model-assisted auditing, and final adjudication, resulting in 1,170 revised-and-certified items. The remaining 689 items are released as a documented uncertain set with explicit uncertainty sources and expertise tags for future refinement. We evaluate seven state-of-the-art language models on HLE and HLE-Verified, observing an average absolute accuracy gain of 7--10 percentage points on HLE-Verified. The improvement is particularly pronounced on items where the original problem statement and/or reference answer is erroneous, with gains of 30--40 percentage points. Our analyses further reveal a strong association between model confidence and the presence of errors in the problem statement or reference answer, supporting the effectiveness of our revisions. Overall, HLE-Verified improves HLE-style evaluations by reducing annotation noise and enabling more faithful measurement of model capabilities. Data is available at: https://github.com/SKYLENAGE-AI/HLE-Verified

skylenage-ai SKYEVAL
·
Feb 14 3

Evidence-Guided Recovery for AI Agents with Ambiguous Tool Outcomes and Partial Execution

Tool timeouts can leave AI agents facing ambiguous tool outcomes and partial execution: an action may have failed, committed or remained in flight while other workflow steps have already completed. We formulate evidence-guided task recovery over approved obligations, provider capabilities and states compatible with public evidence. A shared admissibility check handles key expiry, authorization timing, dependencies and residual effects; a terminal certificate separates actual goal satisfaction from publicly supported safe completion. The known-model implementation combines standard observation-contingent planning with a separately implemented effect-history adjudicator. In a finite study of 37 assignments, planning and three active controls each certify 28 completions; four saved synthetic fee units incur additional computation. A preregistered assignment-held-out comparison then evaluates planning advice against evidence prompting with two local model bundles. Across 58 paired units in 19 previously developed task families, advice certifies 30 completions versus 9, a 36.2-percentage-point difference (exact family sign-flip p=0.0078), using 272 proposals versus 399. All 116 trajectories pass independent replay. Both models exhibit a positive completion difference. The comparison measures the evaluated advice interface under a common shield; search-versus-rule advice requires a separate matched comparison. Unknown and handoff outcomes remain explicit, with guarantees tied to the declared provider capabilities rather than arbitrary third-party exactly-once execution.

  • 4 authors
·
Oct 29 3

Interlock: From Local Correctness to Task Completion in Long-Horizon Agents

Large language model agents increasingly undertake long-horizon tasks in complex environments characterized by fragmented evidence, partially completed states, and dependencies across resources and artifacts. Existing agents may satisfy individual requirements while failing to maintain these relationships, producing locally correct but incomplete outcomes. These failures arise because task completion depends on relationships among environmental evidence, consequential decisions, and required effects. To address this completion gap, we propose Interlock, a framework for task synthesis that represents each task as a typed completion graph. Grounding edges connect public evidence to decisions, coupling hyperedges impose joint constraints across components, and realization edges connect decisions to all required downstream effects. Interlock compiles this graph into a resettable workspace, a public request, and an evaluation rubric. A source-coverage review checks whether the original requirements survive generation, while a public-input review checks whether the rubric is supported by the materials available to the solver. An agent-as-judge inspects execution artifacts and runs verification scripts, retaining trajectories judged to satisfy all mandatory requirements. To evaluate training utility, we fine-tune two Qwen3.6 backbones on their own accepted trajectories. Across the two backbones, training improves AutomationBench task pass rates by 4.00-4.50 percentage points, JobBench weighted scores by 3.20-6.64 points, and Harvey LAB criterion pass rates by 3.89-4.16 percentage points. These results demonstrate the training utility of the synthesized experience across external task environments.

meituan-longcat LongCat
·
Oct 13 3

RGDT-Bench: Benchmarking LLM Reasoning for Rule-Governed Decisions and Their Justifications

We study reasoning in Rule-Governed Decision Tasks (RGDTs), where models apply external rules to case facts and justify decisions, as required in policy, contract, and compliance settings. Beyond the deductive capability emphasized by standard mathematical and logical reasoning tasks, RGDTs require interpreting rules and their applicability, assessing conditions from evidence, combining judgments under rules and exceptions, and providing checkable justifications. These demands motivate a benchmark assessing both decisions and their stated grounds. We introduce RGDT-Bench, providing 202.1K condition-level supervision slots across four task tracks and eight supported task-probe combinations that vary access to supporting information. Label-blind extraction and deterministic checks produce labels for warrant completeness: source-referenced coverage and consistency of stated decision grounds. The benchmark attributes failures to four process layers: rule use, condition, evidence, and aggregation, and checks the final outcome. Among evaluable correct responses, warrant incompleteness averages 40.2% across six evaluated LLMs and supported task-probe combinations. Such warrant incompleteness poses potential safety risks and remains difficult to detect: the best of seventeen existing evaluators reaches only 57.69% (random: 50%) task-averaged area under the receiver operating characteristic curve (AUROC). To address this difficulty, we train a simple reward model with warrant supervision. It achieves 69.24% task-averaged AUROC among correct answers, exceeding the matched outcome-supervised baseline by 10.37 pp (percentage points) and the best existing evaluator by 11.55 pp. Beyond completeness assessment, the model outperforms both outcome-supervised baselines across nearly all response-selection comparisons, supporting RGDT-Bench's warrant supervision for RGDT reasoning.

  • 10 authors
·
Sep 27

CatchBench: When Can an Agent Failure Be Caught?

When can an agent failure be caught? An audit is usually limited by the record rather than by the method. CatchBench therefore puts one auditor's question to three information states: the declared configuration before a run (PRE), a growing prefix of its trace (LIVE), and the finished trace (POST). Prior benchmarks fix one of these states or vary the telemetry; to our knowledge none scores all three under one task-method interface. Each state admits different questions, so seven task contracts carry their own labels and metrics rather than one leaderboard. Four are evidential; three are Gold-derived mechanism diagnostics. The release scores 72 entrants, from rule scanners and structural models to eleven LLM judges across nine model families (GPT, Claude, Gemini, Gemma, Llama, Qwen, DeepSeek, Mistral, Nova), over 1187 declared configurations and 1162 recorded runs. Most of the arena does not order: 56 of 138 registered contrasts separate, and the rest are published unresolved rather than ranked. The two sharpest results cut against our own data. One rule ignores every name and permission; it flags each capability declared after the first. On one of six configuration sources it reaches a perfect F1, so a score there measures how the corpus was built rather than how well a method reasons. Our admissibility bar then rejected one injected substrate and withheld evidential status from the other. A benchmark number is therefore not interpretable until the process behind its labels is published and tested for the shortcut it may leave. We report both, and regenerate every ordering from released predictions with no model call.

  • 5 authors
·
Sep 5

When No Benchmark Exists: Validating Comparative LLM Safety Scoring Without Ground-Truth Labels

Many deployments must compare candidate language models for safety before a labeled benchmark exists for the relevant language, sector, or regulatory regime. We formalize this setting as benchmarkless comparative safety scoring and specify the contract under which a scenario-based audit can be interpreted as deployment evidence. Scores are valid only under a fixed scenario pack, rubric, auditor, judge, sampling configuration, and rerun budget. Because no labels are available, we replace ground-truth agreement with an instrumental-validity chain: responsiveness to a controlled safe-versus-abliterated contrast, dominance of target-driven variance over auditor and judge artifacts, and stability across reruns. We instantiate the chain in SimpleAudit, a local-first scoring instrument, and validate it on a Norwegian safety pack. Safe and abliterated targets separate with AUROC values between 0.89 and 1.00, target identity is the dominant variance component (η^2 approx 0.52), and severity profiles stabilize by ten reruns. Applying the same chain to Petri shows that it admits both tools. The substantial differences arise upstream of the chain, in claim-contract enforcement and deployment fit. A Norwegian public-sector procurement case comparing Borealis and Gemma 3 demonstrates the resulting evidence in practice: the safer model depends on scenario category and risk measure. Consequently, scores, matched deltas, critical rates, uncertainty, and the auditor and judge used must be reported together rather than collapsed into a single ranking.

A Controlled Candidate-Set Benchmark for Offline Satellite-Security Plan Decomposition

Some security-development settings require local models that map an objective to an ordered, checkable plan. We present a low-rank decomposition adapter and release a case-disjoint corpus with 24 authored decompositions and 83 derived next-step examples across 24 satellite-security cases. To prevent reference-plan leakage, prompts contain one objective and eight shuffled SPARTA candidates. The evaluation candidate sets are constructed with oracle inclusion of every reference technique, so this study measures selection from a controlled set, not retrieval. On six fixed cases and five decoding seeds, mean scores vary across adapters and prompted Qwen2.5 baselines; a separate greedy check over three adapter training seeds shows non-negligible training variance. A 24-fold greedy leave-one-case-out control at the two compact sizes likewise shows tradeoffs: at 1.5B, adapter precision is 0.583 against 0.480 for two-shot prompting, while recall is 0.586 against 0.660. Case-resampling intervals for adapter precision relative to the strongest prompted baseline cross zero at 1.5B and 7B. The adapter is also substantially more format-compliant, which prevents attributing all score differences to selection. Autoregressive next-step accuracy is 0.08, 0.06, and 0.29 at 0.5B, 1.5B, and 7B. We therefore offer the formulation, leakage-controlled dataset, reproducibility package, and descriptive analysis as a proof of concept, not evidence of a reliable stand-alone decomposition system or a general adapter advantage. A separate diagnostic tests contract reuse.

  • 3 authors
·
Jul 28

HANDBOOK.md: A Benchmark for Long-Context Agentic Instruction Following

Language-model agents are increasingly deployed under standing instructions: a system prompt, a policy file, or a skills document is placed in context, and the agent is trusted to let that document govern every action that follows. Existing benchmarks rarely test this deployment pattern directly; they measure whether an agent can complete a task, not whether a long, binding policy document constrains its behavior over an extended tool-use horizon. We present HANDBOOK_md, a benchmark of 65 agentic tasks modeled on how employees follow company handbooks. Each task places an agent in a self-contained company environment (a file workspace with mock email, chat, calendar, issue-tracking, and commerce services exposed over the Model Context Protocol) and instructs it to carry out routine professional work governed by an expert-written standard operating procedure of 20-124 pages. Tasks span five domains (finance, medical billing, insurance, logistics, and HR) and 10 fictional companies. To resist memorization, every task modifies one of 10 base handbooks, altering the specific rules and thresholds on which grading depends, so no two tasks share the same set of policies. Grading is fully deterministic: each task carries a rubric of programmatic criteria (824 in total) that check both that required actions occurred and that prohibited actions did not. Under strict grading, where a trial passes only if every criterion is satisfied, the strongest evaluated model passes 36.2% of trials, and most frontier models remain below 25%. Failures follow consistent patterns: agents let a plausible but unauthorized in-environment request override the standing policy, perform a required check and then act against its result, lose rule details over long horizons, and report compliance they did not achieve. We release the tasks, environments, and evaluation harness.

  • 7 authors
·
Aug 2

KWBench: Measuring Unprompted Problem Recognition in Knowledge Work

We introduce the first version of KWBench (Knowledge Work Bench), a benchmark for unprompted problem recognition in large language models: can an LLM identify a professional scenario before attempting to solve it. Existing frontier benchmarks have saturated, and most knowledge-work evaluations to date reduce to extraction or task completion against a specification. KWBench targets the step before that: recognizing the governing structure of the situation from raw inputs alone. The benchmark contains 223 tasks sourced from practitioners across acquisitions, contract negotiations, clinical pharmacy, organizational politics, fraud analysis, and incentive design. Each task encodes a formal game-theoretic pattern (principal-agent conflict, signaling, mechanism design failure, strategic omission, coalitional dynamics, strategic interdependence) and carries structured ground truth recording the expert reading of the situation and the anticipated failure modes. Models receive raw data and a task prompt with no indication of problem type. Scoring is a three-tier rubric gated by a mandatory conjunctive check. Mandatory criteria encode the predicted wrong paths. We evaluate 16 models. The best model passes on 27.9% of tasks. The top two models agree on only 31.7% of their passes. Among the top 8, 44 tasks are solved by exactly one model; routing across the top 8 covers 50.7% of the benchmark, nearly double the best single model. Conditional on passing, quality scores converge (approx 83% across models); unconditional scores do not. Same models articulate the relevant game-theoretic concept correctly when asked, then fail to apply it unprompted. We release KWBench to shift how frontier models are evaluated on knowledge work, scoring them on whether they recognize the right problem from the situation alone, not only on how well they execute once the problem has been framed for them.

clio-ai Clio AI
·
Apr 16 2

Responsible Task Automation: Empowering Large Language Models as Responsible Task Automators

The recent success of Large Language Models (LLMs) signifies an impressive stride towards artificial general intelligence. They have shown a promising prospect in automatically completing tasks upon user instructions, functioning as brain-like coordinators. The associated risks will be revealed as we delegate an increasing number of tasks to machines for automated completion. A big question emerges: how can we make machines behave responsibly when helping humans automate tasks as personal copilots? In this paper, we explore this question in depth from the perspectives of feasibility, completeness and security. In specific, we present Responsible Task Automation (ResponsibleTA) as a fundamental framework to facilitate responsible collaboration between LLM-based coordinators and executors for task automation with three empowered capabilities: 1) predicting the feasibility of the commands for executors; 2) verifying the completeness of executors; 3) enhancing the security (e.g., the protection of users' privacy). We further propose and compare two paradigms for implementing the first two capabilities. One is to leverage the generic knowledge of LLMs themselves via prompt engineering while the other is to adopt domain-specific learnable models. Moreover, we introduce a local memory mechanism for achieving the third capability. We evaluate our proposed ResponsibleTA on UI task automation and hope it could bring more attentions to ensuring LLMs more responsible in diverse scenarios. The research project homepage is at https://task-automation-research.github.io/responsible_task_automation.

  • 4 authors
·
Jun 1, 2023

Self-Evaluating Recursive Agents

Recursive language-model agents decompose tasks and delegate subtasks to child instances of the same policy, forming a tree of work. Training them, however, is hard: the final outcome is verifiable, but the self-invented intermediate subtasks are numerous and carry no ground truth. Existing methods score each node with a verifier or judge, which is costly at scale and blind to decomposition quality. We argue that a recursive agent must learn three coupled capabilities within one set of weights: decomposing problems into subtasks, solving them, and evaluating the outcomes, each requiring its own training signal. SERA (Self-Evaluating Recursive Agents) turns evaluation into a learned capability of the policy itself. Before delegating, the parent writes a rubric of weighted success criteria for each child subtask; a ranking objective against verified outcomes then trains rubric generation so that the criteria track genuine subtask success. In addition, a complementary leaf-coverage signal provides direct credit for task decomposition. Our central finding is that training the policy to generate aligned rubrics is what drives the gains: because the same weights both evaluate and execute, learning to judge subtasks sharpens the agent's ability to solve them. Notably, external supervision is also reduced: the judge is consulted only to train the rubric generator, while solving is trained against the agent's own rubric scores, which outperform direct use of the judge. Beyond training, the learned rubric doubles as an inference-time selector for tree search. On TextCraft-Synth and TextWorld-Sync, SERA improves over strong recursive-agent baselines by 5.38 and 13.14 points on average, and rubric-guided tree search at inference adds a further 2.43 points on TextWorld-Sync.

Bounded Reachability & Jailbreak Detection via Contraction-Constrained State Space Models

Safety heads are lightweight classifiers attached to pretrained language models for flagging harmful inputs before generation. Their empirical detection performance has been studied, but their formal robustness properties remain largely unexplored. We ask when a State Space Model (SSM)-based safety head can be certified to produce the same prediction for all inputs within a bounded embedding-space perturbation. We prove that the answer turns on a single condition: the l_infty norm of the state transition matrix must satisfy A_infty<1 (the contraction condition), which enables exact interval bound propagation (IBP) certification for linear time-invariant classifiers. When the contraction condition holds, the reachable output interval has bounded steady-state width and examples can be certified as robustly classified. When it fails, the interval grows exponentially with sequence length and certification is impossible at any practical perturbation radius. We enforce contraction with a hinge penalty and show on toxic comment data that certified fraction improves from 41\% to 59\%, with a sharp empirical phase transition at A_infty=1 matching the theory. Applying a contraction-regularized S4 head to jailbreak detection on JailbreakBench, we achieve a zero-shot transfer to AdvBench (DR=0.994) and HarmBench (DR=0.988). A logistic regression on mean-pooled Mamba-130M embeddings matches or exceeds the S4 head on every detection metric, confirming that harmful intent is already linearly separable in the embedding space. The S4 safety head's contribution is not superior discrimination but the formal certification that no probe-based approach provides.

  • 1 authors
·
Oct 1

LLM-as-a-Verifier: A General-Purpose Verification Framework

Scaling pre-training, post-training, and test-time compute have become the central paradigms for improving the capabilities of LLMs. In this work, we identify verification, the ability to determine the correctness of a solution, as a new scaling axis. To unlock this and demonstrate its effectiveness, we introduce LLM-as-a-Verifier, a general-purpose verification framework that provides fine-grained feedback for agentic tasks without requiring additional training. Unlike standard LM judges that prompt LLMs to produce discrete scores for candidate solutions, LLM-as-a-Verifier computes the expectation over the distribution of scoring token logits to generate continuous scores. This probabilistic formulation enables verification to scale along multiple dimensions: (1) score granularity, (2) repeated evaluation, and (3) criteria decomposition. In particular, we show that scaling the scoring granularity leads to better separation between positive and negative solutions, resulting in more calibrated comparisons. Moreover, scaling repeated evaluation and criteria decomposition consistently lead to additional gains in verification accuracy through variance and complexity reduction. We further introduce a cost-efficient ranking algorithm for selecting the best solution among candidates using the verifier's continuous scores. LLM-as-a-Verifier achieves state-of-the-art performance on Terminal-Bench V2 (86.5%), SWE-Bench Verified (78.2%), RoboRewardBench (87.4%), and MedAgentBench (73.3%). Beyond verification, the fine-grained signals from LLM-as-a-Verifier can also serve as a proxy for estimating task progress. We build an extension for Claude Code, enabling developers to monitor and improve their own agentic systems. Finally, we show that LLM-as-a-Verifier can provide dense feedback for RL, improving the sample efficiency of SAC and GRPO on robotics and mathematical reasoning benchmarks.

  • 9 authors
·
Jul 5 1

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis

Working agents need to read diverse files, coordinate tools, and produce deliverables. Training such agents requires tasks built on many real files with verifiable results, but few pipelines exist to synthesize this kind of data. Existing pipelines either generate files with models, which lack realism and diversity, or build tasks on real files without task-specific verifiers, leaving result quality unchecked. We introduce GraphForge, an evidence-graph based framework that grounds both the task and its verification in real files. Starting from occupation-grounded seeds for controlled diversity, GraphForge assembles a workspace of real files for each seed and builds an evidence graph over their relations. Since the task statement and rubrics are both derived from this graph, task requirements are backed by the workspace files and each criterion is anchored to the files needed to verify it. An initial rollout further tests executability, and a revision agent repairs the task and rubrics against the original files before trajectories are collected. Fine-tuning Qwen3.6-27B on 2,169 GraphForge trajectories brings GDPVal to 1445.7 (+65.7) under OpenHands, and Workspace-Bench-Lite and SpreadsheetBench II to 63.7 (+7.7) and 24.0 (+13.7) under Claude Code. Rejection fine-tuning on the SFT model's own rollouts, with candidates selected by the evidence-anchored rubrics, yields further improvements on all three benchmarks, suggesting that the rubrics provide a useful selection signal. The data and models are available.

VerifierBench: Benchmarking the Generation of Executable Verifiers for LLM Agents

Agents need reliable ways to validate their outputs, both when completing tasks and when filtering synthetic training data. We study verifier generation: from the instruction and environment available for solving a task, an agent writes an executable verifier that must accept every valid solution and reject invalid ones. We introduce VerifierBench, a benchmark of 222 tasks from six agent benchmarks spanning tool-call, state-based, and code verification, with 4,993 hidden test cases that include alternative valid solutions and attacks; Verifier4Verifier (V4V), a six-check audit, screens its source verifiers and test suites. Across 19 models in a common agent harness, normalized balanced accuracy ranges from 10.6 to 62.1, with generated verifiers consistently rejecting invalid solutions more reliably than they accept valid ones. This apparent strength can mask overly restrictive verification: most false rejections in the analyzed records come from verifiers that accept no tested solution to a task. Case studies reveal unsupported assumptions about solution formats, tool use, and code changes that exclude valid solutions. These errors survive self-testing because agents test solutions that share the verifier's mistaken assumptions, or submit verifiers without confirming that any correct solution passes. These findings suggest testing both whether a verifier accepts a correct solution and whether it accepts other solutions permitted by the task. The project page and dataset are publicly available.

  • 7 authors
·
Oct 18 2

Making Agents More Consistent: Skills Should Form Habits for Repeat Tasks

On repeated work, agents are inconsistent. We ran 42 tasks three times each and found that, depending on the model, 38% to 74% returned answers that did not agree. Consistency is what a buyer, an auditor, or a regulator requires, and agents do not have it. They are wasteful too: 95.3% to 97.2% of what an agent generates goes to re-deriving a plan the system already knows. We propose skill habit formation. An agent mines its own execution history for candidate skills, deterministic variants that compete against the incumbent rather than replacing it. A candidate declares the region of input space it claims, so the common case runs as a script and the rest falls through to reasoning. Four gates of ascending cost admit candidates; the central one tests a candidate's execution trace against a retained reference, within a tolerance measured from that reference's own run-to-run variability. On text-to-SQL, three of four reasoning arms reproduced their own output on 11 to 13 of 42 repeated questions and the fourth on 26 of 42, while a habit-formed variant reproduced on all 456 dispatches we repeated and was non-inferior to every arm it replaced (p<0.0001). It also used 14% to 56% fewer tokens, turning net positive after 7 to 53 reuses. We measured what this costs in accuracy. The guard admitted work it should have deferred on 2.6% of natural paraphrases and 26% of inputs near its boundary, and 11 of 13 such failures were invisible to the trace-conformance gate at any threshold. Deterministic errors repeat exactly: a bad habit is as reliable as a good one, and that is the price of the property that makes the system auditable. Separating routing from parameter extraction raised end-to-end accuracy from 0.888 to 0.952 at 43% of the cost.

  • 2 authors
·
Sep 20

ScientistOne: Towards Human-Level Autonomous Research via Chain-of-Evidence

Autonomous research agents produce competitive solutions and professional-looking manuscripts, yet their outputs contain verifiability failures undetectable by surface-level evaluation: fabricated citations, unreproducible scores, and method descriptions that diverge from the implementation. We address this through three contributions. First, Chain-of-Evidence (CoE), a verifiability framework requiring every claim to be traceable to its evidence source. Second, ScientistOne, an end-to-end autonomous research system that maintains evidence chains by construction throughout literature review, solution discovery, and paper writing. Third, CoE Audit, a post-hoc audit whose four integrity checks -- score verification, specification violation, reference verification, and method-code alignment -- apply uniformly to all systems. Across 75 papers spanning five systems and five frontier research tasks, every baseline exhibits at least one systematic failure mode: hallucinated reference rates reach 21%, score verification passes in as few as 42% of papers, and method-code alignment ranges from 20% to 80%. ScientistOne achieves zero hallucinated references (0/337), perfect score verification (12/12), and the highest method-code alignment (14/15), while matching or exceeding human expert performance on all five tasks. ScientistOne further generalizes to six additional tasks spanning medical imaging, fine-grained recognition, 3D perception, and language modeling, achieving state-of-the-art on Parameter Golf and gold medals on MLE-Bench tasks where baselines fail entirely.

google Google
·
May 24 2

SURE-RAG: Sufficiency and Uncertainty-Aware Evidence Verification for Selective Retrieval-Augmented Generation

Retrieval-augmented generation (RAG) grounds answers in retrieved passages, yet relevance does not guarantee sufficiency: a topical passage may still fail to justify the answer. We study evidence sufficiency verification for selective RAG answering, in which a verifier receives a question, a candidate answer, and retrieved evidence and decides whether the evidence supports, refutes, or is insufficient for the answer, answering only when support is established. We present SURE-RAG, an aggregation protocol that treats evidence sufficiency as a set-level property: missing hops and unresolved conflicts cannot be detected by scoring passages independently. A shared claim-evidence verifier produces a local relation distribution for each (claim, passage) pair, which SURE-RAG aggregates into four interpretable answer-level feature blocks (coverage, relation strength, uncertainty, and retrieval), producing a three-way decision and an auditable selective score. We evaluate on HotpotQA-RAG v3, a controlled multi-hop benchmark, under an artifact-aware protocol (shortcut baselines, counterfactual swaps, no-oracle checks, and GPT-4o audits). Calibrated SURE-RAG attains 0.9075 Macro-F1 (raw 0.8951 +/- 0.0069), well above DeBERTa mean-pooling (0.6516) and a GPT-4o judge (0.7284), and on par with a strong concat cross-encoder (0.8888 +/- 0.0109) while remaining fully auditable. At 30% coverage, risk falls from 0.2588 to 0.1642, a 37% relative reduction. As a boundary-mapping experiment, we contrast SURE-RAG with GPT-4o on HaluBench unsafe detection: the ranking reverses (0.3343 vs. 0.7389 unsafe-F1), indicating that controlled sufficiency verification and natural hallucination detection are distinct problems.

  • 3 authors
·
Jul 23

StealthBench: Measuring Operational Stealth in Autonomous Offensive-Security Agents

Stealth, the discipline of achieving an objective without revealing your presence, capabilities, or collected intelligence, is what separates sophisticated operators from detectable ones. Elite security researchers and advanced persistent threats achieve their objectives unnoticed; autonomous agents increasingly inherit the same offensive tasks, but do they inherit the tradecraft? We introduce StealthBench,a benchmark that measures operational stealth in autonomous offensive-security agents across six operational security (OPSEC) dimensions. We extract 11 hand-verified OPSEC incidents from real bug-bounty and red-team trajectories, expanded into 14 dockerized task scenarios, where agents, despite finding real vulnerabilities, committed stealth failures inconsistent with standard operational tradecraft: embedding credentials in public uploads, deleting production resources to prove access, force-adding uninvolved users to demonstrate a race condition. We evaluate agent trajectories using a 3-model large language model (LLM) judge panel with majority-vote aggregation, measuring safe success rate (solved and stealthy), Stealth@Solve (tradecraft quality among successful solves), and reckless solve rate (solved but cover blown). Our results show that no model exceeds 54% safe success rate (the compound metric requiring both task completion and stealth), confirming that OPSEC failures are systematic across model families. We release StealthBench as a public benchmark to support both the development of stealth-aware agents and automated OPSEC monitoring for autonomous offensive-security deployments. The interactive leaderboard, evaluation harness, and dataset are available at https://stealthbench.com.

  • 2 authors
·
Jul 27 2

Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning

Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes. We study this as a deployment allocation problem rather than a new-verifier problem. We introduce \sevra, Selective Verification for Reasoning Allocation, a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification. Using a frozen Qwen3-4B solver, we log intervention outcomes and train recoverability-aware gates from serving-visible attempt state. On \mathfive, selective verification reaches 76.3\% accuracy, compared with 75.5\% for always verifying, while reducing post-generation tokens by 26.8\% and harmful flips from 2.2\% to 1.0\%. However, an 8,192-token initial solve reaches 76.0\% accuracy with 28\% fewer total model tokens, showing that selective recovery is useful but not the best tested cost frontier. In frozen transfer to \gsm, the selective policy verifies only 3.0\% of examples, improves accuracy from 93.4\% to 94.5\%, and reduces verification tokens by 91.2\% relative to always verifying; again, a longer initial solve matches its accuracy with fewer realized tokens. On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy at about five times the realized token cost. The resulting deployment rule is: tune the initial budget first, then use selective recovery when explicit checks, bounded retries, auditability, or regression-risk control matter.

IndustryBench: Probing the Industrial Knowledge Boundaries of LLMs

In industrial procurement, an LLM answer is useful only if it survives a standards check: recommended material must match operating condition, every parameter must respect a regulated threshold, and no procedure may contradict a safety clause. Partial correctness can mask safety-critical contradictions that aggregate LLM benchmarks rarely capture. We introduce IndustryBench, a 2,049-item benchmark for industrial procurement QA in Chinese, grounded in Chinese national standards (GB/T) and structured industrial product records, organized by seven capability dimensions, ten industry categories, and panel-derived difficulty tiers, with item-aligned English, Russian, and Vietnamese renderings. Our construction pipeline rejects 70.3% of LLM-generated candidates at a search-based external-verification stage, calibrating how unreliable industrial QA remains after LLM-only filtering.Our evaluation decouples raw correctness, scored by a Qwen3-Max judge validated at κ_w = 0.798 against a domain expert, from a separate safety-violation (SV) check against source texts. Across 17 models in Chinese and an 8-model intersection over four languages, we find: (i) the best system reaches only 2.083 on the 0--3 rubric, leaving substantial headroom; (ii) Standards & Terminology is the most persistent capability weakness and survives item-aligned translation; (iii) extended reasoning lowers safety-adjusted scores for 12 of 13 models, primarily by introducing unsupported safety-critical details into longer final answers; and (iv) safety-violation rates reshuffle the leaderboard -- GPT-5.4 climbs from rank 6 to rank 3 after SV adjustment, while Kimi-k2.5-1T-A32B drops seven positions.Industrial LLM evaluation therefore requires source-grounded, safety-aware diagnosis rather than aggregate accuracy. We release IndustryBench with all prompts, scoring scripts, and dataset documentation.

CARVE: Certified Affordable Repair of Vetoed Maneuvers via Envelopes for Interactive Driving

Interactive driving exposes a failure mode that is easy to miss in rule-aware autonomous-driving stacks: a hard-rule margin can be negative for an ego candidate even though a small lawful accommodation by a non-priority agent would restore feasibility. Existing rulebooks, shields, and reachability filters are strong at vetoing unsafe actions, while prediction-based planners model likely responses. Neither returns a runtime proof object that states which bounded multi-agent edit repairs the maneuver, who owns the edit, whether the request is right-of-way affordable, and what ego fallback remains if the request is not observed. We formulate this missing object as *interactive repair certification* and introduce *CARVE*, a prediction-free certificate layer over a finite lattice of ego-owned and agent-owned tactical operators. Agent-owned requests are admissible only inside \(B_j(s) = β(π_j)α_j^{\max}(s)\), a cooperation envelope that separates kinematic reachability from normative priority. The resulting certificate records the binding rule, repair category, repair set, responsibility-weighted cost split, and fallback. On 589 Lanelet2-geometry-grounded INTERACTION replay episodes, CARVE-Greedy accepts 98.64% of initially vetoed maneuvers and recovers 370/378 human-resolved false vetoes, while preserving 589/589 right-of-way respect, zero priority-agent false positives, and 400/400 negative-stress vetoes. We prove certificate soundness, structural right-of-way respect, exact finite-lattice minimality, fallback contingency, and blame-consistency conditions. CARVE does not predict or require another driver's compliance; it certifies whether a proposed interaction is bounded, attributable, and normatively admissible under declared assumptions.

  • 1 authors
·
May 30 2

Theoria: Rewrite-Acceptability Verification over Informal Reasoning States

When should an AI system's answer be trusted? Formal proof assistants offer certainty but cannot reach most of the problem distribution; scalar LLM judges offer coverage but produce opaque scores that cannot be audited after the fact and are subject to the same coherence issues as any LLM. We present Theoria, a verification architecture that closes this gap. A candidate solution is rewritten into a sequence of typed state transitions, each licensed by an explicit justification, whether that be a citation, computation, or problem-given fact, and every transition is independently auditable. The foundational invariant is completeness of change: every difference between consecutive proof states must be accounted for, so hidden premises surface as unlicensed mutations rather than passing silently. On HLE-Verified Gold (185 text-only expert problems), Theoria certifies 105 at 91.4% strict precision (Wilson 95% CI [84.5%, 95.4%]). Every certification produces a human readable proof trace in which each step can be independently challenged. Holistic LLM judges achieve comparable precision at matched coverage but fail on different problems (Jaccard 0.14-0.36), making the approaches complementary. On 95 adversarial poisoned proofs across 15 domains, structured judges catch 94.7% versus 83.2% for holistic judging (p= 0.0017). The overall 11.5 pp gap concentrates in hidden premises (90.6% vs. 62.5%, a 28 pp difference) and fabricated citations (100% vs. 90%), the error classes where the formal analysis predicts an advantage; performance is identical on arithmetic and theorem-misapplication errors, where no advantage is predicted. On GPQA Diamond (n= 65), certified precision is 97.1% (Wilson CI [85.1%, 99.5%]).

  • 2 authors
·
Jul 1

SkillHorizon: Constructing Long-Horizon Agent Tasks through Skill Composition

Long-horizon agents must complete workflows in which later activities depend on earlier outputs and changes to shared state. A key challenge for evaluation is to expand task requirements and their dependencies while keeping tasks executable and outcomes verifiable. We introduce SkillHorizon, a benchmark that constructs long-horizon tasks through skill composition in seven business environments. Its generators use domain-specific rules to turn reusable skill procedures into concrete business activities. These activities are connected through input-output dependencies, shared resources, and state updates, allowing workload and coordination demands to grow together. Successful reference executions validate task feasibility and measure interaction workload, while programmatic evaluators assess local completion and whole-task success. We evaluate five language models on 1,050 tasks with reference interaction workloads ranging from 64K to 1M tokens. Across these scales, mean local-goal completion falls from 90.4% to 46.8%, as completed work fails to keep pace with growing requirements. At larger workloads, agents more often submit unsuccessful final answers with substantial work unfinished and tool-call allowance remaining. Trace analysis also identifies cases where scripts execute successfully but their outputs fail downstream business checks. SkillHorizon provides a configurable testbed for studying sustained progress, dependency coordination, and complete task delivery in growing workflows. Our code is available at https://github.com/KnowComp/SkillHorizon.

  • 14 authors
·
Oct 9

Teaching Agents to Code Reliably

Autonomous coding agents solve repository issues by reading code, running commands, editing files, and submitting patches. Extra inference-time compute yields gains only when it produces a useful repair and supplies reliable evidence for choosing one. Three behaviors decide both, and we argue they are teachable rather than byproducts of scale, so a policy can carry them instead of a scaffold. Location diversity remains narrow, since attempts return to the same site and extra samples add no coverage. Edit diversity is left unexploited, since methodologies that differ resolve complementary issues no single run reaches. Verification misleads, since a test the agent writes for its own patch accepts many incorrect ones. Directing search by execution feedback and scoring each patch against its own reverted tree resolves 52.8% of SWE-bench Verified using 48.1% of the agent-steps an eight-sample baseline spends. Training moves these behaviors into the policy. On the 270 issues held out from SFT and RL training, weighted supervised fine-tuning raises pass@1 from 31.9% to 35.2% and pass@8 from 46.7% to 51.1%. A reinforcement objective then trains the verifier against gold-labeled repairs and incorrect variants, crediting the assertions that detect them. It raises pass@1 to 43.0% and pass@8 to 60.7%, lifts verifier precision from 26.8% to 41.7%, and more than halves false acceptance. Resolution improves on two of three out-of-distribution suites and verifier precision on all three, and the gains hold at 7B, 14B, and 30B against published coder baselines.

  • 6 authors
·
Oct 1

Scores Alone Do Not Prove Discovery: The Discovery Certification Protocol for Auditing AI Research Agents

AI research agents combine prior knowledge, public sources, and experimental feedback to produce useful results. The Discovery Certification Protocol (DCP) turns claims about these results into executable recovery and feedback tests. Gate 1 validates useful improvement on sealed evaluation. Gate 2 gives matched agents the registered starting information and observed Web content while withholding the target research history. Every valid method reaching the numerical target supplies a recovery witness and triggers the Core veto. DCP Core requires adequate controls, zero observed recoveries, and a finite-sample bound on recovery in one fresh registered episode. Optional Gate 3 measures the average effect of truthful feedback relative to a specified neutral policy from a shared checkpoint. DCP Evidence adds this effect after independent null calibration and a registered effect margin. Two controlled audits exercise the complete protocol in SQLite optimization and virtual catalyst control under different models. Each produced zero recoveries in 96 episodes, with an upper bound of 0.0468. Each paired study yielded 30 truthful recoveries and zero neutral recoveries, with passing 60-pair null studies. Additional cases exercise Core, recovered, and audit-incomplete decisions. A deterministic, LLM-free verifier reproduces the decisions from frozen evidence. DCP provides a common evidence language for useful outcomes, alternative routes, and feedback effects across AI research.

WorkflowArena: Benchmarking Executable Workflow Orchestration in LLM-Based Multi-Agent Systems

Large language model agents are increasingly deployed as multi-agent systems rather than as a standalone model operating within a single context. Recent agent harnesses allow the model to instantiate worker agents and to author the executable workflow that coordinates them, shifting the design of system structure from developers to the model at run time. The quality of this organization therefore constitutes a distinct capability, yet existing benchmarks either measure end-to-end task completion or evaluate coordination under structures specified in advance. We formalize this capability as workflow orchestration and introduce WorkflowArena, a benchmark of 145 long-horizon tasks spanning 8 professional domains, from software engineering and formal mathematics to legal analysis and 3D production. Each task specifies an initial workspace and an integrated deliverable composed of interdependent components, which is assessed by a hidden verifier. Agents determine autonomously whether and how to delegate, and all delegation is recorded, enabling joint evaluation of deliverable quality and the organization of the underlying work. Experiments with nine models show that no single configuration performs best across all domains and that systems exhibit distinct delegation strategies: at high effort, Claude models in Claude Code invoke workflows less frequently but with larger worker groups, whereas GPT models in Codex delegate more frequently in smaller units. When the executing model is held fixed on a challenging subset of tasks, workflows authored by Claude Opus 5.5 substantially improve weaker executors, indicating that workflow organization is itself a measurable determinant of task performance.

  • 13 authors
·
Nov 8

CANDOR: Chance-Calibrated Neighborhood Discordance in Frozen Encoders for Medical Imaging

A foundation encoder is pretrained once on a large image corpus and then reused with its weights frozen. Each new task is solved by training a small head on the features it produces. This setup is common in medical imaging, where labeled cases are scarce and a frozen encoder can be reused across findings. All downstream tasks then depend on the class separation present in that fixed feature space. Encoder selection usually uses the area under the receiver operating characteristic curve (AUROC) of a trained downstream head. AUROC measures the predictive information a head can extract from the features, but it does not measure class separation in the frozen feature space. A positive image that lies near a negative image in feature space has a bounded normalized margin under any Lipschitz head. Chance-calibrated neighborhood discordance (CANDOR) measures this feature-space separation without training a head. For a positive image, it compares the k nearest positive-label neighbors with the k nearest negative-label neighbors, among images acquired the same way. Discordance rate D is the share of positives whose opposite-label neighbors are nearer. Drawing the 2 candidate sets at equal size makes the labels interchangeable, so label-independent features have chance level D=1/2 without simulation. We apply CANDOR to 22 frozen encoders on 605,443 images from 20 public datasets, covering 8 binary tasks in 7 domains. On every task, the best encoder is below 1/2. A discordant image bounds the normalized margin of every Lipschitz head on that encoder. A selector that is shown the true label and chooses among 11 encoders reduces the miss rate from 0.359 to 0.028. Discordance is associated with occlusion retention and with none of pretraining objective, parameter count, release year, or finding size. The fixed chance level lets D be computed for a frozen encoder before any downstream head is trained.

  • 3 authors
·
Sep 19

Signed but Not Safe: Requirement Integrity in Multi-Agent LLM Handoffs

Enterprise AI agents increasingly split a user's request across subagents and merge their results, and each handoff can summarize away, alter or ignore a requirement. We report a preregistered study with synthetic, exactly scorable tasks comparing free-text, structured JSON and signed requirement-manifest handoffs: the orchestrator extracts each requirement as a verbatim quote plus a checkable predicate, code signs the manifest with HMAC-SHA256, and deterministic code verifies the merger's decision. With Qwen3-8B (1,440 trajectories, five threats), signed manifests cut unsafe commits under attack from 69.4% to 9.0% (60.4 pp, 95% CI [51.4, 69.4]) with no net loss of conformant commits relative to structured handoffs. Signing is what defeats in-transit tampering: an unsigned manifest committed violations in 79% of mutated runs, the signed one in none. On identical trajectories the deterministic verifier caught 78.7% of violating decisions versus 59.0% for an LLM checker, with 7% versus 50% false alarms. An oracle manifest removed all remaining unsafe commits: the residual risk comes from extraction errors made before signing. A preregistered pre-signing injection attack confirmed this boundary: one adversarial line raised signed-manifest violations from 14.6% to 37.5% (Qwen3-8B), 11.1% to 63.9% (gpt-oss-20b) and 8.3% to 75.0% (GPT-5.6-luna), and the signature then authenticated the corrupted manifest. Signing protects requirements in transit, not at extraction; the extraction step is the trust boundary.

Gym-Anything: Turn any Software into an Agent Environment

Computer-use agents hold the promise of assisting in a wide range of digital economic activities. However, current research has largely focused on short-horizon tasks over a limited set of software with limited economic value, such as basic e-commerce and OS-configuration tasks. A key reason is that creating environments for complex software requires significant time and human effort, and therefore does not scale. To address this, we introduce Gym-Anything, a framework for converting any software into an interactive computer-use environment. We frame environment creation itself as a multi-agent task: a coding agent writes setup scripts, downloads real-world data, and configures the software, while producing evidence of correct setup. An independent audit agent then verifies evidence for the environment setup against a quality checklist. Using a taxonomy of economically valuable occupations grounded in U.S. GDP data, we apply this pipeline to 200 software applications with broad occupational coverage. The result is CUA-World, a collection of over 10K long-horizon tasks spanning domains from medical science and astronomy to engineering and enterprise systems, each configured with realistic data along with train and test splits. CUA-World also includes CUA-World-Long, a challenging long-horizon benchmark with tasks often requiring over 500 steps, far exceeding existing benchmarks. Distilling successful trajectories from the training split into a 2B vision-language model outperforms models 2times its size. We also apply the same auditing principle at test time: a separate VLM reviews completed trajectories and provides feedback on what remains, improving Gemini-3-Flash on CUA-World-Long from 11.5% to 14.0%. We release all code, infrastructure, and benchmark data to facilitate future research in realistic computer-use agents.

  • 3 authors
·
Apr 6

Agent Skills Can Be Harmful: An Empirical Study of Skill-Induced Failures in LLM Agents

Agent skills are the de facto mechanism for extending LLM agents with reusable guidance. A skill can shape the agent's task execution, including planning, tool use, problem-solving, and validation. Prior work reported mixed results of agent skills: some skills improve task success rates, while others have no effect, increase token use and execution time, and even reduce success rates. This paper presents a comprehensive analysis of skill-induced agent failures by attributing task failures and cost regressions to specific loaded skills. We introduce a differential analysis framework that attributes a failure or regression to a skill by comparing a target skill-guided run against a no-skill or semantically matched skill reference run that solves the same task, or solves it more cheaply. We instantiate this framework on SkillsBench and SWE-Skills-Bench, yielding 307 skill-induced failures, including 125 functional failures and 182 efficiency regressions. We also build SkillTriage, a taxonomy-guided attribution tool that normalizes paired cases, extracts differential evidence, and produces triage reports. Our major findings include: (1) Skill induced functional failures are rarely caused by obviously irrelevant skills; instead, seemingly relevant skills often make the agent incorrectly implement or omit task-required implementation elements. (2) Skill-induced efficiency regressions are not explained by prompt length alone. (3) The largest sources within Excessive Procedure are excessive verification and heavy implementation pipelines, contributing 67 and 30 cases, respectively. This shows that skills often turn validation checklists and construction recipes into mandatory work. Based on our findings, we propose research topics and tooling improvements for safer and more cost-aware skill reuse.

  • 6 authors
·
Aug 11

Time Travel in LLMs: Tracing Data Contamination in Large Language Models

Data contamination, i.e., the presence of test data from downstream tasks in the training data of large language models (LLMs), is a potential major issue in measuring LLMs' real effectiveness on other tasks. We propose a straightforward yet effective method for identifying data contamination within LLMs. At its core, our approach starts by identifying potential contamination at the instance level; using this information, our approach then assesses wider contamination at the partition level. To estimate contamination of individual instances, we employ "guided instruction:" a prompt consisting of the dataset name, partition type, and the random-length initial segment of a reference instance, asking the LLM to complete it. An instance is flagged as contaminated if the LLM's output either exactly or nearly matches the latter segment of the reference. To understand if an entire partition is contaminated, we propose two ideas. The first idea marks a dataset partition as contaminated if the average overlap score with the reference instances (as measured by ROUGE-L or BLEURT) is statistically significantly better with the completions from guided instruction compared to a "general instruction" that does not include the dataset and partition name. The second idea marks a dataset partition as contaminated if a classifier based on GPT-4 with few-shot in-context learning prompt marks multiple generated completions as exact/near-exact matches of the corresponding reference instances. Our best method achieves an accuracy between 92% and 100% in detecting if an LLM is contaminated with seven datasets, containing train and test/validation partitions, when contrasted with manual evaluation by human experts. Further, our findings indicate that GPT-4 is contaminated with AG News, WNLI, and XSum datasets.

  • 2 authors
·
Aug 16, 2023

Semantic Overlays: Mitigating Prompt Injection with Annotations Beyond Tokens and Steering Vectors

Everything a language model sees is tokens. The serving stack knows what each span is -- user input, tool output, instructions -- but the model must keep track of that itself, and it can lose track or be confused: text can be written to read like anything. Prompt injection is a natural exploit of this phenomenon. By scrambling the model's understanding of span identity, an attacker can induce unwanted and potentially dangerous actions. Adding a non-textual channel to the model's input -- a way to communicate span identity beyond text -- mitigates this class of attack. We thus introduce a general steering technique called Semantic Overlays: small learned adapters applied at chosen prefill positions to a frozen model's residual stream. Laying an overlay over a span creates an out-of-band annotation channel that cannot be replicated by tokens. Unlike steering vectors, Semantic Overlays are trained, adaptable, and selectively applied. An overlay can encode complex semantics that reshape how the model perceives the marked span: asked to copy a code snippet under an overlay asserting that it is in a different programming language than it is, the model rewrites the snippet, faithfully, in the asserted language. Overlays are also composable, allow for transparent reading of underlying content, and can carry complex payloads -- including imperatives that the model will follow. An overlay which marks a span as "non-executable" defends against the broad class of prompt injections that add instructions in untrusted context. We report strong results on prompt injection benchmarks: SEP separation rises from 24.3% to 96.5% with utility unchanged (our scoring rule; we also correct a defect in the published grader), TensorTrust attack success rate falls from 34.8% to 6.6%, and all four PIArena attack families drop to 0% compliance, all while marked spans stay readable (92.5% exact copy rate).

  • 1 authors
·
Aug 23

Evaluating Bounded Autonomy in Regulated Agentic AI: A Diagnostic Harness with Constitutional Rewards, Escalation Labels, and Runtime Governance

We propose RegLLM, a diagnostic harness for bounded autonomy in regulated agentic workflows. It instruments six trustworthiness signals: citation validity, source grounding, schema compliance, escalation correctness, constitutional alignment, and unsafe-action rate. Signals are distinguished by their source of supervision: programmatic verifiers, task-level escalation labels, or AI-judge scores. A deterministic runtime supervisor blocks ungrounded answers and forces escalation, logging interventions. The same domain constitution informs evaluation, training rewards, and serving guardrails. Task-level should-escalate labels make the act-versus-defer decision a measurable training signal. We demonstrate the harness at smoke scale. An offline reference run (n=12) lifts escalation recall from 0 to 0.67 and reduces unsafe-action rate from 0.33 to 0.08 when governance is enabled. Two single-GPU Qwen2.5-3B LoRA/DPO pilots (n=8, same seed and evaluation split) expose substantial variation: nominally identical RL-base configurations yield task success of 0.25 versus 0.12 and escalation recall of 1.0 versus 0.5. An answer-quality adapter changes recall from 1.0 to 0.5 in Run A, but from 0.5 to 1.0 in Run B. An escalation-aware variant produces no measurable change in Run B. These small pilots do not establish reliable adapter effects or production readiness. Their contribution is diagnostic: configuration variance can overwhelm apparent tuning effects on bounded-autonomy metrics, motivating larger evaluation sets and repeated runs.

  • 1 authors
·
Sep 27 1

The Verifier Is Part of the Simulator: Evidence-View Gaming in Agentic Equation Discovery

A growing number of agentic pipelines search for governing equations: an agent proposes laws and a verifier checks them against simulated trajectories. The verifier sees only a numerical summary: integrals against test functions on a time grid. That summary carries its own discretization error, making the verifier part of the simulator. We freeze the trajectory, candidate library, verifier and thresholds, leaving the agent one move: which permitted summary to admit. Without scores or language models, finer time grids reduce numerical discrepancies and false library rejections. On PDEBench, the verifier instead accepts equations missing diffusion on score-selected summaries. Maximizing a public score rewarding decisive verdicts and larger mismatches when rejecting the library wrongly rejects all 1,500 in-library cases; retrospectively retuning score constants on a separate bank reduces errors to 13/180 at best. We call this evidence-view gaming: reward hacking by choice of permitted summary. In single-turn forced-choice tests, Claude models rarely choose summaries giving wrong verdicts when asked to be accurate, but often do so when asked to maximize the score or report missing physics. In separate trust prompts, they often rank those harmful summaries least trustworthy. Requiring several summaries to agree reduces errors but often withholds a verdict and performs worse than a fixed fine view on PDEBench: a safeguard, not a certificate. Evidence chosen after its verdict is known cannot alone authorize a claim.

  • 2 authors
·
Oct 31

TessPay: Verify-then-Pay Infrastructure for Trusted Agentic Commerce

The global economy is entering the era of Agentic Commerce, where autonomous agents can discover services, negotiate prices, and transact value. However adoption towards agentic commerce faces a foundational trust gap: current systems are built for direct human interactions rather than agent-driven operations. It lacks core primitives across three critical stages of agentic transactions. First, Task Delegation lacks means to translate user intent into defined scopes, discover appropriate agents, and securely authorize actions. Second, Payment Settlement for tasks is processed before execution, lacking verifiable evidence to validate the agent's work. Third, Audit Mechanisms fail to capture the full transaction lifecycle, preventing clear accountability for disputes. While emerging standards address fragments of this trust gap, there still remains a critical need for a unified infrastructure that binds the entire transaction lifecycle. To resolve this gap, we introduce TessPay, a unified infrastructure that replaces implicit trust with a 'Verify-then-Pay' architecture. It is a two plane architecture separating control and verification from settlement. TessPay operationalizes trust across four distinct stages: Before execution, agents are anchored in a canonical registry and user intent is captured as verifiable mandates, enabling stakeholder accountability. During execution, funds are locked in escrow while the agent executes the task and generates cryptographic evidence (TLS Notary, TEE etc.) to support Proof of Task Execution (PoTE). At settlement, the system verifies this evidence and releases funds only when the PoTE satisfies verification predicates; modular rail adapters ensure this PoTE-gated escrow remains chain-agnostic across heterogeneous payment rails. After settlement, TessPay preserves a tamper-evident audit trail to enable clear accountability for dispute resolution.

  • 3 authors
·
Jan 29 1

OSWorld2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks

Existing computer-use benchmarks fail to capture the realism, complexity, and long-horizon demands of real-world computer use, limiting their ability to reveal the limitations of frontier agents. We introduce OSWorld 2.0, a benchmark of 108 long-horizon computer-use workflows across everyday and professional tasks, designed to capture complex and challenging real-world phenomena. Each task represents a realistic end-to-end workflow that takes human users a median of about 1.6 hours to complete and requires an average of 318 tool calls with Claude Opus 4.7 using maximum thinking, compared with about 30 in OSWorld 1.0. OSWorld 2.0 targets challenge phenomena that are common in real workflows yet underrepresented in prior benchmarks, spanning interaction-design challenges such as streaming interaction and dynamic environments, as well as agent-pattern challenges such as cross-source reasoning, implicit-state inference, and visual-spatial precision. Tasks are grounded in authentic input artifacts and cross-referenced against realistic stateful user profile data, and include separate safety reports auditing safety-sensitive execution. Under our primary binary-completion metric at 500 steps, Claude Opus 4.8 with maximum thinking and batched tool calls scores best but still completes only 20.6% of tasks at a 54.8% partial score; GPT-5.5 is far more token-efficient yet plateaus near 13%. These results show that current agents are still far from professional-level computer use: rather than stumbling on basic GUI control or coding, they lose track of constraints, miss information that arrives mid-task, guess rather than ask the user, and skip verification, struggling most when a task hinges on hidden state they must recover.

  • 36 authors
·
Jun 27 1

VibeLifeBench: Can Your Life Agent Be Proactive and Persistent in a Living World?

Large language model (LLM) agents are increasingly deployed as personal assistants. Existing evaluations, however, mostly use short, self-contained requests in static environments. Everyday life assistance is different. A task runs for weeks rather than minutes. The world keeps changing while the agent is not being prompted. Many constraints are never stated outright. An agent that merely answers the request in front of it will fail at such a task. What is needed instead is an agent that stays proactive and consistent. It decides on its own when to act, when to ask, and when to stay silent. It notices changes that nobody announced. It keeps one plan coherent from the first day to the last. No current benchmark measures this. We introduce VibeLifeBench, a benchmark of 200 long-horizon tasks across ten everyday-life domains. Each task is a scripted multi-week timeline in a simulated world of 22 mock services. The world advances on its own clock, and many of its changes are silent, so only an agent that re-inspects the world discovers them. Every task is graded by fine-grained, weighted checks that read only what the agent actually left behind, covering the end state, the timeliness of its actions, and whether it upheld the implicit constraints. We evaluate seven frontier models. All of them score low, which shows how far current agents are from assisting with real life. We will open-source all tasks, environments, and the evaluation framework.

dots-studio dots studio
·
Aug 10 2

From Answers to States: Verifiable Process-Level Evaluation of Chemical Reasoning in Large Language Models

Large language models are increasingly used as chemistry assistants, yet most chemistry benchmarks still score only final answers. This masks a critical failure mode: a model may output the correct molecule, product, or option while its reasoning violates chemical logic. Existing process-level evaluators are hard to scale because LLM judges and human step-level process annotation are costly, inconsistent, and vulnerable to hallucination. We introduce ChemCoTBench-V2, a rule-verifiable diagnostic benchmark for low-cost, auditable evaluation of structured, verifier-addressable chemical reasoning traces. It spans molecular understanding, molecule editing, molecular optimization, and reaction prediction, with 5,620 evaluation samples across 18 reporting tasks. Models must expose key intermediate steps in expert-designed templates, and those steps are checked with deterministic chemistry rules and, for closed-answer tasks, reference traces rather than another LLM judge. Open-ended molecular optimization is evaluated with oracle-verifiable state constraints rather than strict trace matching. The benchmark reports three separate signals: final-answer correctness, template adherence, and step-wise verifier correctness over expert-refined intermediate commitments. Experiments on frontier models reveal a persistent gap between final-answer success and structured-reasoning-state consistency: models often follow the requested format while failing chemical-step checks, or answer correctly with weak supporting reasoning. ChemCoTBench-V2 enables fine-grained model comparison and identifies the concrete step at which the trace first violates the verifier.

  • 5 authors
·
Jun 2

MedSkillAudit: A Domain-Specific Audit Framework for Medical Research Agent Skills

Background: Agent skills are increasingly deployed as modular, reusable capability units in AI agent systems. Medical research agent skills require safeguards beyond general-purpose evaluation, including scientific integrity, methodological validity, reproducibility, and boundary safety. This study developed and preliminarily evaluated a domain-specific audit framework for medical research agent skills, with a focus on reliability against expert review. Methods: We developed MedSkillAudit (skill-auditor@1.0), a layered framework assessing skill release readiness before deployment. We evaluated 75 skills across five medical research categories (15 per category). Two experts independently assigned a quality score (0-100), an ordinal release disposition (Production Ready / Limited Release / Beta Only / Reject), and a high-risk failure flag. System-expert agreement was quantified using ICC(2,1) and linearly weighted Cohen's kappa, benchmarked against the human inter-rater baseline. Results: The mean consensus quality score was 72.4 (SD = 13.0); 57.3% of skills fell below the Limited Release threshold. MedSkillAudit achieved ICC(2,1) = 0.449 (95% CI: 0.250-0.610), exceeding the human inter-rater ICC of 0.300. System-consensus score divergence (SD = 9.5) was smaller than inter-expert divergence (SD = 12.4), with no directional bias (Wilcoxon p = 0.613). Protocol Design showed the strongest category-level agreement (ICC = 0.551); Academic Writing showed a negative ICC (-0.567), reflecting a structural rubric-expert mismatch. Conclusions: Domain-specific pre-deployment audit may provide a practical foundation for governing medical research agent skills, complementing general-purpose quality checks with structured audit workflows tailored to scientific use cases.

AIPOCH-AI AIPOCH
·
Apr 21 2

Long SKILL Compliance as Logical Reasoning: Closure-Grounded Detection with Scaling-Guided On-Policy Distillation

The increasing complexity of enterprise business scenarios has promoted the widespread adoption of long SKILL documents in agent systems, posing new challenges for compliance detection: large models incur substantial inference costs, while small models may fail to maintain detection accuracy. To address this gap, we propose SkillCDG, a graph-based framework for long SKILL compliance detection. SkillCDG represents complex business policies as a two-layer constraint dependency graph, where the upper layer indexes SKILL descriptions for scenario routing and the lower layer captures dependencies among atomic constraints within each SKILL. During inference, two-level retrieval followed by dependency closure supports compliance judgment and source traceability. We comprehensively evaluate the framework on three enterprise datasets and two controlled public benchmark variants. Experimental results demonstrate that SkillCDG outperforms baseline methods by up to 12.8 percentage points in detection F1 score, while reducing token consumption by a maximum 64.3\%. Moreover, we further investigate the inherent relationships among policy-graph complexity, model scale, and detection performance. Comparative experiments conducted on four checkpoints from a single model family validate a concise and effective scaling trend: end-to-end detection correctness exhibits a complexity-differentiated scaling pattern, and the complexity metric derived from the constraint dependency graph can effectively quantify instance difficulty and the performance improvement potential of models. Leveraging this insightful scaling trend, we conduct adaptive training sample selection and adopt on-policy distillation to efficiently enhance the compliance detection capability of small-scale models.

  • 7 authors
·
Aug 7

Rethinking Benchmark and Contamination for Language Models with Rephrased Samples

Large language models are increasingly trained on all the data ever produced by humans. Many have raised concerns about the trustworthiness of public benchmarks due to potential contamination in pre-training or fine-tuning datasets. While most data decontamination efforts apply string matching (e.g., n-gram overlap) to remove benchmark data, we show that these methods are insufficient, and simple variations of test data (e.g., paraphrasing, translation) can easily bypass these decontamination measures. Furthermore, we demonstrate that if such variation of test data is not eliminated, a 13B model can easily overfit a test benchmark and achieve drastically high performance, on par with GPT-4. We validate such observations in widely used benchmarks such as MMLU, GSK8k, and HumanEval. To address this growing risk, we propose a stronger LLM-based decontamination method and apply it to widely used pre-training and fine-tuning datasets, revealing significant previously unknown test overlap. For example, in pre-training sets such as RedPajama-Data-1T and StarCoder-Data, we identified that 8-18\% of the HumanEval benchmark overlaps. Interestingly, we also find such contamination in synthetic dataset generated by GPT-3.5/4, suggesting a potential risk of unintentional contamination. We urge the community to adopt stronger decontamination approaches when using public benchmarks. Moreover, we call for the community to actively develop fresh one-time exams to evaluate models accurately. Our decontamination tool is publicly available at https://github.com/lm-sys/llm-decontaminator.

  • 5 authors
·
Nov 8, 2023 1

Benign in Isolation, Harmful in Composition: Security Risks in Agent Skill Ecosystems

Skills are becoming the capability layer through which LLM agents turn plans into actions, but their use introduces security risks such as data leakage, unauthorized operations, and tool misuse. Existing vetting usually evaluates each skill in isolation, while real agent tasks often invoke multiple skills in a shared execution context. This creates Skill Composition Risk (SCR): a skill that appears benign alone can become harmful when its outputs, trust signals, authorization cues, or side effects influence later invocations along an activated path. We introduce SCR-Bench to evaluate this risk in controlled, sandboxed skill environments. Rather than relying only on textual intent or surface behavior, SCR-Bench records downstream state changes and path-level outcomes across composed skill executions. It contains three sub-benchmarks: SCR-CapFlow for capability-flow composition, SCR-TrustLift for trust-transfer composition, and SCR-AuthBlur for authorization-confusion composition. Across SCR-Bench, composed paths expose risks that are largely absent under isolated evaluation. In SCR-CapFlow, attack success rate reaches 33.6 percent under composition, compared with near-zero isolated baselines. In SCR-TrustLift, attack success rate exceeds 96.5 percent on four of five backends. In SCR-AuthBlur, the risky-approval rate increases by 71.8 percent relative to the L0 isolated baseline under the L1 context setting. These results show that agent skill security should be assessed at the level of activated paths rather than isolated artifacts. SCR and SCR-Bench provide a foundation for path-aware risk evaluation and defense in LLM agent skill ecosystems. Benchmark: https://github.com/saint-viperx/SCR_Bench.

  • 5 authors
·
Jun 12

Phantom Success, Silent Failure: Auditing the Success Checkers of Simulated Manipulation Benchmarks

Simulated manipulation benchmarks measure policy performance using hand-written success checkers that label each episode as a success or failure. Passing an expert demonstration shows that a checker accepts one way of completing the task. Learned policies, however, may stop after partial progress or complete the task in a different way. A checker may then accept an incomplete execution (phantom success) or reject an execution that satisfies the instruction (silent failure). In this paper, we introduce SimAudit, an LLM-based pipeline that audits success checkers against task instructions. SimAudit analyzes the instruction and checker code to identify scenarios where the two may disagree, recreates those scenarios in simulation, and runs the original checker. Multi-view renderings are then used to verify whether the observed behavior satisfies the instruction. Across RoboTwin 2.0, RoboCasa365, and RoboDojo, 183 simulated test cases produced the predicted checker verdict. Visual verification and deduplication yielded 61 distinct defects across 49 tasks, including 40 phantom successes and 21 silent failures. Most defects in RoboCasa365 accepted incomplete executions, most in RoboDojo rejected valid executions, and both error types were common in RoboTwin 2.0. These findings suggest that validating success checkers requires testing both incomplete executions and valid alternatives to expert demonstrations, and motivate more rigorous evaluation protocols for future benchmarks.

Goedel-Code-Prover: Hierarchical Proof Search for Open State-of-the-Art Code Verification

Large language models (LLMs) can generate plausible code but offer limited guarantees of correctness. Formally verifying that implementations satisfy specifications requires constructing machine-checkable proofs, a task that remains beyond current automation. We propose a hierarchical proof search framework for automated code verification in Lean~4 that decomposes complex verification goals into structurally simpler subgoals before attempting tactic-level proving. Central to our approach is a principled decomposition score that combines constructive justification with structural effectiveness. Crucially, this score serves as both the training reward and the inference-time ranking criterion, ensuring strict alignment between optimization and deployment. We train Goedel-Code-Prover-8B, a single unified policy for both decomposition and completion, via supervised initialization followed by hybrid reinforcement learning, where a continuous decomposition reward drives planning exploration while supervised replay stabilizes proof generation. On three Lean-based code verification benchmarks comprising 427 tasks, our 8B-parameter model achieves a 62.0\% prove success rate, a 2.6times improvement over the strongest baseline, surpassing neural provers up to 84times larger. We further observe consistent inference-time scaling: success rates improve monotonically with search iterations and sampling budget, with our trained model achieving greater efficiency than frontier off-the-shelf models of comparable scale.

  • 11 authors
·
Mar 18

Crypto Accounting Bench: Evaluating Frontier and Open-Weight Models on Crypto-Asset Accounting Tasks

We introduce Crypto Accounting Bench (CAB), a benchmark for assessing whether frontier and open-weight language models can reconstruct the complete journal entry that an organization actually posted for a crypto-asset transaction. CAB contains 118 evaluation tasks drawn from 7 pseudonymized organizations. Each task combines transaction mechanics, asset quantities and base-currency values, wallet and legal-entity context, counterparty evidence, related transaction legs, recurrence, tax-lot evidence, and the organization's complete chart of accounts. The target is a balanced structured entry with every required account, side, amount, currency, and full-precision asset quantity. We evaluate 12 models spanning proprietary frontier systems and open-weight releases over 3 independent attempts per task, producing 4,248 trajectories. We report 3 metrics: Mean Score, Best@3, and Pass@3. Pass@3 is the fraction of tasks with at least 1 of 3 attempts that satisfies every rubric criterion and required gate. The leading model reaches 77.43% Mean Score, while the best Pass@3 is 56.78%. Deterministic diagnostics, read from each task's best of 3 attempts and macro-averaged across the 12 models, show higher base-amount agreement (97.8%) than deciding-account accuracy (56.3%). Together with the failure analysis, these results identify account selection and complete-entry composition as the main remaining challenges on CAB.

  • 3 authors
·
Sep 12

Building Safe and Reliable AI systems for Safety Critical Tasks with Vision-Language Processing

Although AI systems have been applied in various fields and achieved impressive performance, their safety and reliability are still a big concern. This is especially important for safety-critical tasks. One shared characteristic of these critical tasks is their risk sensitivity, where small mistakes can cause big consequences and even endanger life. There are several factors that could be guidelines for the successful deployment of AI systems in sensitive tasks: (i) failure detection and out-of-distribution (OOD) detection; (ii) overfitting identification; (iii) uncertainty quantification for predictions; (iv) robustness to data perturbations. These factors are also challenges of current AI systems, which are major blocks for building safe and reliable AI. Specifically, the current AI algorithms are unable to identify common causes for failure detection. Furthermore, additional techniques are required to quantify the quality of predictions. All these contribute to inaccurate uncertainty quantification, which lowers trust in predictions. Hence obtaining accurate model uncertainty quantification and its further improvement are challenging. To address these issues, many techniques have been proposed, such as regularization methods and learning strategies. As vision and language are the most typical data type and have many open source benchmark datasets, this thesis will focus on vision-language data processing for tasks like classification, image captioning, and vision question answering. In this thesis, we aim to build a safeguard by further developing current techniques to ensure the accurate model uncertainty for safety-critical tasks.

  • 1 authors
·
Aug 6, 2023

FrontierChallenge: Evaluating Scientific Workflow Completion

Scientific agents increasingly analyze data, execute code, and produce research artifacts, yet most benchmarks emphasize final answers, isolated programs, or a single domain. We introduce FrontierChallenge, a cross-domain benchmark comprising 300 end-to-end scientific workflows. In this paper, we release and evaluate 97 of these tasks, spanning quantum chemistry, molecular dynamics, materials characterization, analytical chemistry, life science, and electrochemistry/environment. Each task provides fixed inputs and specifies a bundle of required scientific deliverables. We evaluate twelve frontier models with three agent scaffolds. Pass Rate measures the fraction of tasks satisfying the full-completion criterion, while Avg. Score captures partial progress. Each of the best-performing configurations completed only 20 of the 97 released tasks, yielding a Pass Rate of 20.6%. Partial progress translated especially poorly into complete delivery in analytical chemistry and electrochemistry/environment: Avg. Scores reached 87.6 and 94.9, but the highest Pass Rates were only 4% and 0%. Among non-passing Claude Code trajectories, 75.5% still ended with language claiming completion. These findings show that neither high partial scores nor confident claims of completion reliably indicate that a scientific task has been fully delivered, highlighting the need to evaluate end-to-end workflow execution and the completeness of scientific deliverables together.

apodex Apodex
·
Aug 24 2

DeliverBench: Benchmarking AI Agents on Delivering Realistic Multi-Stage Professional Projects

LLM agents are judged by what they deliver, and in professional work what gets delivered is a project, not a task. A project pays off only if the whole package holds, and one weak part can limit the value of the rest. Yet existing benchmarks mostly test agents on one task at a time, leaving a gap between the capability we measure and the outcome that makes agents useful. To address this gap, we introduce DeliverBench, which tests whether agents can carry a project from brief to final delivery. Its 11 projects, spanning financial due diligence, legal contract drafting, and drug safety reporting, are reconstructed with 13 domain experts from real materials and chain 150 subtasks through 2-15 rounds of new materials and client feedback. Because no reference outputs are injected, agents inherit their own errors from one task to the next. Expert rubrics then score every checkpoint as well as the final package. Across seven model-harness systems, even the best system scores only 65.0 out of 100 on final delivery and passes 50.0% of subtasks at a 90-point threshold. The failures are often failures of carrying work forward: in one audit project, all seven systems addressed the review findings, but only one preserved traceability. Together, these results suggest that task-level capability does not simply compose into delivery, which also depends on keeping the work coherent as it evolves.

  • 24 authors
·
Nov 2

FRAUDSkill: Structured Frozen-Weight Skill Optimization for Audio Anti-Fraud Detection

Large audio-language models have shown promise for anti-fraud detection by directly processing speech and reasoning over fraud-related evidence. Their deployment, however, requires predictions to follow a predefined label space and a structured decision protocol consisting of service-scenario identification, fraud detection, and conditional fraud-type classification. Existing fine-tuning and prompt-based approaches typically encode task knowledge, constraints, and decision rules into model parameters or manually maintained prompts, making them difficult to adapt as fraud patterns and labeling policies evolve. To this end, we propose FRAUDSkill, a structured frozen-weight adaptation framework that leaves the underlying audio-language model unchanged while optimizing an external layer of skill programs, route-specific policies, and decision rules. We further combine structured output control with validation-guided multi-path inference to ensure protocol-compliant predictions. On the TeleAntiFraud benchmark, FRAUDSkill achieves 73.50% Macro-F1, outperforming the shared frozen-model baseline by 31.96% while reducing invalid outputs to 1.94%. Extensive experiments demonstrate that external skill optimization provides an effective and adaptable solution for structured audio anti-fraud detection without modifying the underlying model. The source code is available at https://anonymous.4open.science/r/FRAUDSKILL-114514.

AstroBench: Benchmarking Free-Floating Capture and Space Robotic Manipulation

Capturing a loose object inside a spacecraft requires a robot to approach a target that can translate and rotate, and whose motion changes when contact is made. These conditions motivate evaluation beyond manipulation of supported objects. We introduce AstroBench, an open benchmark for free-floating capture and space robotic manipulation with a fixed-base arm. The task suite combines four initial-motion tracks with sample-storage operations based on MELFI hardware and cap and captive-screw servicing based on the Robotic Refueling Mission. The two workcells provide nine task definitions and 29 task-target combinations. A metadata-filtered Objaverse inventory supplies 8,414 object instances for asset expansion, alongside five cabin visual contexts and versioned task instructions. We report two acquisition-focused experiments using existing checkpoints. A frozen π_{0.5} DROID joint-position policy captures 2 of 200 static targets and 4 of 200 moving targets across twenty object identities. A cube-specific ACT policy captures 8 of 10 static targets and 6 of 10 moving targets in a separate experiment. Each experiment retains its recorded observation, action and capture configuration. The results document difficult direct transfer for the tested generalist checkpoint and successful task-specific capture behavior, while identifying concrete settings for studying motion-aware perception and contact-sensitive control. Code, task definitions and editable workcell assets are available at https://github.com/AstroUpLab/AstroBench.

  • 6 authors
·
Oct 21

SCX Router: Streaming Zero-Shot Model Selection with a Decoder-KV Classifier and a Real-World Task Ontology

The rapid proliferation of large language models (LLMs) and the growing diversity of their applications presents a unique optimization opportunity: selecting the right model for the task, while optimizing for speed, cost, and quality at a per-task level. However, inference endpoints can vary widely in quality, price, latency, context support, tool use, domain expertise, and reasoning behavior. This heterogeneity makes manual heuristics difficult to maintain and unlikely to achieve consistently favorable speed--cost--quality trade-offs on their own. We introduce , a lightweight GLiClass-based router that assigns a suitability score to each inference-time model label without autoregressive generation. The released 0.6B-parameter checkpoint combines a Qwen3 decoder with a shallow bidirectional scorer. Its decoder-KV execution path preserves a text-only key--value cache across a session, encodes only new dialogue turns, and evaluates transient candidate-label tokens without adding them to the persistent cache. The same checkpoint also predicts task type, difficulty, reasoning mode, and expected output length, and supports custom zero-shot labels. For task generation, we construct a task ontology with 23 families, 115 task types, 345 routable subtypes, 1,173 synthetic examples, and an orthogonal axis of 30 domains. Using this structure, we generate 150,000 verifier-scored tasks and 15,000 open-ended tasks. We then train the Qwen3 decoder on these tasks, while explicitly separating learned request prediction from per-task policies for attributes such as eligibility, cost, cache reuse, safety, and sovereignty. Across six LiveBench subsets, the router outperforms the mean candidate; on the selected 1,000-task subset, it achieves an aggregate top-1 score of 0.707 versus 0.696 for the strongest fixed model, with benchmark-dependent gains.

  • 5 authors
·
Sep 1

Safe Actions Alone Do Not Ensure Safe Agents: Identifying Unfulfilled Obligations with Guard Models

Guard models are increasingly used to safeguard LLM-based agents, primarily by identifying actions that agents are forbidden to perform. However, identifying forbidden actions alone is insufficient to ensure agent safety. In this paper, we argue that agent safety also depends on identifying required yet unperformed safety-critical actions, which we call obligations. Our preliminary study on a popular benchmark for evaluating safety shows that 56.92% of GLM-5.3 trajectories contain unfulfilled obligations, compared with only 30.00% containing forbidden actions. This finding reveals unfulfilled obligations as a major and previously overlooked source of safety risk. However, to our knowledge, no existing benchmark evaluates whether guard models can identify these obligations. To close this gap, we introduce ObligationBench, the first benchmark for evaluating the capability of obligation identification, comprising 240 expert-validated trajectories covering issue resolution, feature development, and terminal operations. Our evaluation of 14 representative models reveals substantial limitations: the highest recall and exact-match rate are only 48.97% and 10.00%, respectively. To address these limitations, we develop ObligationGuard using 40,000 synthetic training examples. ObligationGuard achieves 57.52% recall and an exact-match rate of 21.67%, surpassing all evaluated models on both metrics. We call on the community to incorporate obligation identification into the design and evaluation of future guard models to improve agent safety.

  • 4 authors
·
Oct 7

Verification Without Sufficiency: Per-Chunk Filtering Fails on Multi-Hop RAG, and Decomposition Repairs It

Verification for retrieval-augmented generation usually scores each retrieved chunk and drops the ones that fail. We show this cannot work for multi-hop questions, and show what does. Per-chunk scoring assumes one chunk is a sufficient premise for the answer. Multi-hop questions are built so that none is, and the paragraph carrying the answer is the one the question does not name. Entailment scoring reaches 0.643, 0.523 and 0.560 AUC on HotpotQA, 2WikiMultihopQA and MuSiQue, against 0.951 on single-hop SQuAD. Seven controls rule out model capacity, premise length, hypothesis template, decision threshold, retriever, answer-matching criterion and prompt. End to end across three datasets, three generator sizes and two prompts, per-chunk gating is significantly worse than not filtering at all in every cell, and its penalty grows with generator capability. The repair is to condition verification on the decomposed sub-question rather than the original query. Using MuSiQue's gold decomposition, entailment on a later hop rises from 0.546, which is chance, to 0.840, a paired lift of +0.355 with a bootstrap interval of [0.331, 0.382]. An off-the-shelf Qwen2.5-7B decomposer, given the question and the top retrieved paragraph, reaches 0.637 and captures 31% of that ceiling; decomposing without retrieval reaches 0.533, below the original question. Iterative retrieval systems already produce such decompositions and discard them before verifying.

  • 1 authors
·
Jul 31

Certifiers Make Neural Networks Vulnerable to Availability Attacks

To achieve reliable, robust, and safe AI systems, it is vital to implement fallback strategies when AI predictions cannot be trusted. Certifiers for neural networks are a reliable way to check the robustness of these predictions. They guarantee for some predictions that a certain class of manipulations or attacks could not have changed the outcome. For the remaining predictions without guarantees, the method abstains from making a prediction, and a fallback strategy needs to be invoked, which typically incurs additional costs, can require a human operator, or even fail to provide any prediction. While this is a key concept towards safe and secure AI, we show for the first time that this approach comes with its own security risks, as such fallback strategies can be deliberately triggered by an adversary. In addition to naturally occurring abstains for some inputs and perturbations, the adversary can use training-time attacks to deliberately trigger the fallback with high probability. This transfers the main system load onto the fallback, reducing the overall system's integrity and/or availability. We design two novel availability attacks, which show the practical relevance of these threats. For example, adding 1% poisoned data during training is sufficient to trigger the fallback and hence make the model unavailable for up to 100% of all inputs by inserting the trigger. Our extensive experiments across multiple datasets, model architectures, and certifiers demonstrate the broad applicability of these attacks. An initial investigation into potential defenses shows that current approaches are insufficient to mitigate the issue, highlighting the need for new, specific solutions.

  • 3 authors
·
Aug 25, 2021

DuMateBench: Evaluating Autonomous Agents in Complex Real-World Workflows

Autonomous agents are increasingly adopted to complete complex, multi-tool workflows in real-world settings. However, existing benchmarks typically separate tasks by application or capability and evaluate agents in environments that are cleaner and more stable than those encountered in practice. We introduce DuMateBench, a real-session benchmark reconstructed from anonymized and privacy-screened user sessions collected from a large-scale production agent platform. Each task preserves the relevant pre-solution interaction history, persistent configurations, and workspace state, and is then validated through human verification. The resulting benchmark comprises 200 tasks spanning 8 broad scenarios and 17 fine-grained capability categories, with most tasks requiring multiple capability coordination. We execute these tasks in isolated Docker containers injected with three forms of real-world environmental complexity: Insufficient, Unstable, and Noisy, and assess performance using a hybrid deterministic and LLM-as-Judge evaluation protocol. Experiments across five representative autonomous-agent frameworks paired with four state-of-the-art LLMs reveal substantial gaps in strict task completion. Complementary robustness, efficiency, and diagnostic analyses further show that performance under environmental perturbations is jointly shaped by the capabilities of the LLM and the surrounding agent framework. The code and data are publicly available at https://dumatebench.com/.

  • 15 authors
·
Aug 26

SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?

We introduce SWE-Bench Pro, a substantially more challenging benchmark that builds upon the best practices of SWE-BENCH [25], but is explicitly designed to capture realistic, complex, enterprise-level problems beyond the scope of SWE-BENCH. SWE-BENCH PRO contains 1,865 problems sourced from a diverse set of 41 actively maintained repositories spanning business applications, B2B services, and developer tools. The benchmark is partitioned into a public set with open access to problems sourced from 11 repositories, a held-out set of 12 repositories and a commercial set of 18 proprietary repositories where we have formal partnership agreements with early-stage startups. Problems in the held-out and the commercial set are not publicly accessible, but we release results on the commercial set. Our benchmark features long-horizon tasks that may require hours to days for a professional software engineer to complete, often involving patches across multiple files and substantial code modifications. All tasks are human-verified and augmented with sufficient context to ensure resolvability. In our evaluation of widely used coding models, under a unified scaffold, we observe that their performance on SWE-Bench PRO remains below 25% (Pass@1), with GPT-5 achieving the highest score to date at 23.3%. To better understand these limitations, we cluster the failure modes observed in the collected agent trajectories for a clearer characterization of the error patterns exhibited by current models. Overall, SWE-BENCH PRO provides a contamination-resistant testbed that more faithfully captures the complexity and diversity of real-world software development, advancing the pursuit of truly autonomous software engineering agents at a professional level.

  • 19 authors
·
Sep 21, 2025 3

ORAgentBench: Can LLM Agents Solve Challenging Operations Research Tasks End to End?

Large language models are increasingly deployed as autonomous agents for multi-step tasks in executable environments, yet their ability to perform realistic operations research (OR) work remains unclear. Existing OR evaluations often decouple modeling from solving, rely on pre-formalized or text-only instances, and rarely test the full workflow from operational artifacts to validated decisions. In this work, we introduce ORAgentBench, an execution-grounded benchmark for evaluating autonomous agents on challenging end-to-end operations research tasks. It contains 107 human-reviewed tasks across diverse operational scenarios, each packaged in an isolated environment with a natural-language brief, multi-file data, configuration artifacts, and a required submission schema. Agents must write and run solution code, and their submissions are evaluated by hidden validators for schema validity, hard-constraint feasibility, and normalized objective quality. Experiments with fourteen frontier agent-model configurations show that current agents remain far from reliable OR practice. The best agent passes only 35.51% of all tasks and 20.59% of hard tasks, and many feasible submissions still fall below the required quality threshold. Failure analysis further shows that errors are dominated by strategic weaknesses, including missed operational rules, brittle formulations, weak feasible-solution construction, and insufficient solution improvement. OR-specific procedural skills increase hard-task feasibility, but do not reliably improve solution quality or pass rate. These results suggest that progress in OR agents requires moving beyond plausible optimization code toward dependable, high-quality operational decision-making.

  • 8 authors
·
Jun 17

Beyond Red-Teaming: Formal Guarantees of LLM Guardrail Classifiers

Guardrail Classifiers defend production language models against harmful behavior, but although results seem promising in testing, they provide no formal guarantees. Providing formal guarantees for such models is hard because "harmful behavior" has no natural specification in a discrete input space: and the standard epsilon-ball properties used in other domains do not carry semantic meaning. We close this gap by shifting verification from the discrete input space to the classifier's pre-activation space, where we define a harmful region as a convex shape enclosing the representations of known harmful prompts. Because the sigmoid classification head is monotonic, certifying the worst-case point is sufficient to certify the entire region, yielding a closed-form soundness proof without approximation in O(d) time. To formally evaluate these classifiers, we propose two constructions of such regions: SVD-aligned hyper-rectangles, which yield exact SAT/UNSAT certificates, and Gaussian Mixture Models, which yield probabilistic certificates over semantically coherent clusters. Applying this framework to three author-trained Guardrail Classifiers on the toxicity domain, every hyper-rectangle configuration returns SAT, exposing verifiable safety holes across all classifiers, despite seemingly high empirical metrics. Probabilistic GMM certificates also expose a divergent structural stability in how these models represent harm. While GPT-2 and Llama-3.1-8B maintain robust coverage of 90% and 80% across varying boundaries, BERT's safety guarantees prove uniquely volatile. This 'coverage collapse' to 55% at the optimal threshold reveals a sparsely populated safety margin in BERT, which only achieves full coverage by adopting an extremely conservative pessimistic threshold. These approaches combined, provide new insights on how effective Guardrail Classifiers really are, beyond traditional red-teaming.

  • 4 authors
·
May 10

PhysicianBench: Evaluating LLM Agents in Real-World EHR Environments

We introduce PhysicianBench, a benchmark for evaluating LLM agents on physician tasks grounded in real clinical setting within electronic health record (EHR) environments. Existing medical agent benchmarks primarily focus on static knowledge recall, single-step atomic actions, or action intent without verifiable execution against the environment. As a result, they fail to capture the long-horizon, composite workflows that characterize real clinical systems. PhysicianBench comprises 100 long-horizon tasks adapted from real consultation cases between primary care and subspecialty physicians, with each task independently reviewed by a separate panel of physicians. Tasks are instantiated in an EHR environment with real patient records and accessed through the same standard APIs used by commercial EHR vendors. Tasks span 21 specialties (e.g., cardiology, endocrinology, oncology, psychiatry) and diverse workflow types (e.g., diagnosis interpretation, medication prescribing, treatment planning), requiring an average of 27 tool calls per task. Solving each task requires retrieving data across encounters, reasoning over heterogeneous clinical information, executing consequential clinical actions, and producing clinical documentation. Each task is decomposed into structured checkpoints (670 in total across the benchmark) capturing distinct stages of completion graded by task-specific scripts with execution-grounded verification. Across 13 proprietary and open-source LLM agents, the best-performing model achieves only 46% success rate (pass@1), while open-source models reach at most 19%, revealing a substantial gap between current agent capabilities and the demands of real-world clinical workflows. PhysicianBench provides a realistic and execution-grounded benchmark for measuring progress toward autonomous clinical agents.

From Specification to Deployment: Empirical Evidence from a W3C VC + DID Trust Infrastructure for Autonomous Agents

Autonomous AI agents now transact at production scale -- 69,000 bots executing 165 million transactions across 50 million USDC in cumulative volume on a single marketplace -- without any shared trust layer between participants. Regulatory frameworks (Singapore IMDA, NIST CAISI, EU AI Act) and major AI laboratories (Anthropic, Google) have independently converged on the same structural requirement: an open, portable, cryptographically verifiable trust infrastructure for autonomous agents that no single vendor can deliver alone. This paper presents MolTrust, a production-deployed implementation of such an infrastructure built on W3C Verifiable Credentials 2.0 and Decentralized Identifiers v1.0, with on-chain anchoring on Base Layer 2. The system architecture is organized around four primitives (identity, authorization, behavioral record, portability), a five-party accountability chain, and the Agent Authorization Envelope (AAE) -- a machine-evaluable authorization structure enforced at three layers: cryptographic signatures, API-level credential lifecycle management, and kernel-level syscall monitoring via Falco eBPF integration. The paper documents three distinguishing capabilities: kernel-layer AAE enforcement below the agent process boundary; cross-protocol interoperability through five reproducible test vectors verified against independent implementations; and layered Sybil resistance combining dual-signature interaction proofs, cross-vertical endorsement diversity gating, and principal-DID-linked violation persistence. The reference implementation has been operational since March 2026 across eight credential verticals. Empirical validation at adversarial scale is pending. The contribution is deployment-first evidence that the trust infrastructure regulators and industry have converged on is implementable today using W3C-standardized primitives.

  • 1 authors
·
May 6

Towards Trustworthy Embodied Intelligence: A Systems Framework and Graded Trustworthiness Levels

Embodied intelligence integrates learned perception and decision making with real-time computation, control, and physical interaction. Because failures can cause immediate physical or operational harm, task completion alone does not establish trustworthiness. We define trustworthy embodied intelligence as the sustained capacity to execute specified tasks reliably under environmental and system variation while maintaining risk within acceptable bounds. We term this objective sustained safe success. Its supporting mechanisms are organized into four interdependent layers. The model layer generates task-competent action proposals with calibrated uncertainty and explicit safety preferences. The system layer realizes authorized actions dependably through integrated sensing, computation, control, hardware safeguards, fault containment, and fallback. The evidence layer substantiates bounded claims through evaluation, verification, validation, traceability, and structured assurance arguments. The deployment layer maintains claim validity through runtime monitoring, authority management, intervention, incident response, and controlled updates. Because assumptions and failures propagate across these layers, neither model capability, isolated safeguards, nor benchmark performance alone can establish end-to-end trustworthiness. Drawing on embodied AI, robotics, control, dependable computing, distributed systems, and autonomous driving, we further propose a non-normative hierarchy of trustworthiness levels. This hierarchy grades the strength of bounded deployment claims across task capability, safety, system assurance, operational governance, and supporting evidence, providing a basis for bounded deployment, comparative evaluation, research prioritization, and future standardization.

  • 41 authors
·
Jul 27

ASI-Bench: At the Dawn of Artificial Superintelligence

Artificial superintelligence (ASI) requires AI to move beyond mastering existing knowledge toward exploring the unknown, creating new knowledge, and turning new ideas into verifiable results. However, the capabilities of today's AI systems are still largely built on learning, compressing, and applying existing human knowledge. Accordingly, existing benchmarks primarily test whether AI can produce correct answers based on learned knowledge, or whether it can complete tasks under extensive human guidance. We therefore introduce ASI-Bench, the first benchmark to jointly evaluate AI systems' capabilities of innovative exploration and autonomous scientific execution across general research domains, and the first to progressively withdraw human methodological guidance within the same research project to test how far AI can proceed on its own. Built by over 40 experts with the cost of 31,000+ human hours, ASI-Bench contains 60 project-level research tasks across 11 scientific domains and progressively reduces methodological guidance to test whether AI can independently select methods, conduct research, and produce verifiable results. All tasks undergo expert review, AI-assisted auditing, sandbox execution, and scorer validation. Across 18 state-of-the-art agent--model configurations, the average score drops from 50.91 with full methodological guidance to 29.10 with only the method specified and 26.62 when agents must determine the method themselves. This sharp decline shows that current systems remain heavily dependent on human guidance and are still far from autonomously conducting end-to-end, project-level scientific research. ASI-Bench is open to the world. We invite researchers and builders everywhere to contribute new tasks, challenge the limits of today's AI, and help accelerate humanity's collective path toward artificial superintelligence at https://asibench.apexin.ai/submit.

  • 42 authors
·
Aug 17 4

FoSeRL: Formal Sequential Robustness Certification for Reinforcement Learning Policies

Even a few action perturbations can substantially degrade the performance of a deployed decision policy. Certifying the resulting return loss is challenging in stochastic environments, where returns vary even without an attack. We introduce FoSeRL, a framework for certifying deployed RL policies against precommitted, temporally sparse action attacks. The deployed policy is unchanged, with no smoothing or retraining. Certification requires a resettable simulator supporting shared randomness and independent one-step successor queries, but no analytical dynamics model. FoSeRL certifies that an attacked episode loses no more than a prescribed amount of return relative to the same episode unattacked, with at least a target probability and at a user-specified confidence level. Both runs share the initial state and randomness, so the measured loss reflects the attack, not the episode; carrying the running return gap as a state coordinate makes it the terminal value, reducing trajectory-level certification to terminal safety. Time-dependent barrier conditions on the augmented state bound the terminal failure probability: satisfied exactly, they certify every admissible precommitted attack; learned from sampled trajectories and verified on held-out data, they certify the same guarantee under a specified attack-episode setting. Across six stochastic continuous-control environments and three RL policy families (TD3, SAC, and PPO), FoSeRL certifies non-trivial cardinality--magnitude robustness frontiers, achieves substantially larger certified budgets than policy smoothing, and reveals marked robustness differences among policies with comparable nominal performance.

  • 4 authors
·
Oct 2

OTAP: Structure-Aware Optimal Transport for Evaluating Planning and Execution in Agent Trajectories

Large language model agents solve tasks by generating trajectories that interleave planning, tool calls, and intermediate results. Current evaluation metrics reduce such a trajectory to a binary success flag, compare it against a reference by exact matching, or delegate judgment to another language model. A success flag cannot distinguish a sound solution from one that succeeds by luck, and says nothing about why a failed run went wrong. Exact matching penalizes plans that are valid but reordered or decomposed differently from the reference. We reframe trajectory evaluation as a distance between the agent's execution graph and a set of valid solution graphs, and instantiate it via an unbalanced fused Gromov-Wasserstein transport problem over attributed dependency graphs. The resulting score, termed OTAP (Optimal Transport for Agentic Planning), is a pseudo-metric that is provably invariant to dependency-preserving reorderings and has bounded sensitivity to redundant steps. Its unbalanced marginals handle missing or hallucinated steps without forcing a match, and its soft coupling accommodates variation in plan granularity. On controlled perturbations and three public benchmarks, OTAP separates valid from invalid trajectories in a regime where semantics-only metrics score below chance. Its advantage tracks the fidelity of the dependency graph: largest where edges follow from operator semantics, smallest where they are inferred from free text. Where a formal verifier exists, strict surface metrics predict validity better than OTAP does, which places OTAP in open-ended domains where no verifier is available.

  • 3 authors
·
Jul 31

Confidently Wrong, Silently So: Auditing Undetectable Failures of a Deployed On-Device Language Model

Aligning deployed language models requires knowing when their outputs can be trusted, yet on-device models now ship to hundreds of millions of devices with no server-side moderation, and the configuration developers can actually deploy is rarely audited independently. We present a reproducible reliability audit of the developer-accessible on-device foundation model, framed as an oversight question: can a user or a resource-constrained developer tell when the model is wrong? Red-teaming it on calibration, confident confabulation on false-premise questions, and over-refusal of benign prompts, we find a task-asymmetric miscalibration: its guardrails fail in opposite directions across tasks (confabulating on 69\% of false premises while refusing 18\% of entirely benign inputs), atop a self-reported confidence that is saturated and non-discriminative (AUROC 0.47; ECE 70, worst among comparable small models). Crucially, confident-correct and confident-wrong outputs are surface-indistinguishable: a classifier over 15 user-visible features separates them at AUROC only 0.55 (equivalence-confirmed), leaving no signal for oversight at inference time. No cheap single-generation signal flags these failures (le0.68 AUROC), whereas a black-box consistency wrapper requiring no model access recovers reliability (confident confabulation 75\%to3\%; selective accuracy 43\%to83\%) at a tunable cost. We contribute a model-agnostic audit protocol, a surface-indistinguishability test, and released code and frozen evaluation items as reusable infrastructure for auditing deployed models.

  • 3 authors
·
Aug 25

Evaluating AI Vocational Skills Through Professional Testing

Using a novel professional certification survey, the study focuses on assessing the vocational skills of two highly cited AI models, GPT-3 and Turbo-GPT3.5. The approach emphasizes the importance of practical readiness over academic performance by examining the models' performances on a benchmark dataset consisting of 1149 professional certifications. This study also includes a comparison with human test scores, providing perspective on the potential of AI models to match or even surpass human performance in professional certifications. GPT-3, even without any fine-tuning or exam preparation, managed to achieve a passing score (over 70% correct) on 39% of the professional certifications. It showcased proficiency in computer-related fields, including cloud and virtualization, business analytics, cybersecurity, network setup and repair, and data analytics. Turbo-GPT3.5, on the other hand, scored a perfect 100% on the highly regarded Offensive Security Certified Professional (OSCP) exam. This model also demonstrated competency in diverse professional fields, such as nursing, licensed counseling, pharmacy, and aviation. Turbo-GPT3.5 exhibited strong performance on customer service tasks, indicating potential use cases in enhancing chatbots for call centers and routine advice services. Both models also scored well on sensory and experience-based tests outside a machine's traditional roles, including wine sommelier, beer tasting, emotional quotient, and body language reading. The study found that OpenAI's model improvement from Babbage to Turbo led to a 60% better performance on the grading scale within a few years. This progress indicates that addressing the current model's limitations could yield an AI capable of passing even the most rigorous professional certifications.

  • 2 authors
·
Dec 16, 2023

Human-Centred, Risk-Constrained Evaluation of Financial Document Recognition: Evidence, Critical Errors, and Selective Automation

Financial document recognition is often assessed using character and word error rates, although these metrics do not express whether a system preserves the financial facts on which people and organisations act. This paper develops a human-centred, risk-constrained framework for evaluating financial document recognition as evidence recovery and decision support. It defines typed financial facts linked to visible source regions and separates verbatim transcription, deterministic normalization, and business interpretation. A layered quality model covers text, localization, layout, tables, fields, fact correctness, evidence grounding, uncertainty, robustness, and operational performance. Non-compensatory acceptance gates use one-sided uncertainty bounds for critical-field errors, preventing strong aggregate scores from masking failures in amounts, identifiers, signs, periods, or entity associations. Selective automation is formulated as a constrained routing problem that accounts for residual risk, human-review capacity, and review cost. The framework further specifies annotation and adjudication procedures, leakage-resistant benchmark partitions, statistical estimands, subgroup analyses, and a reproducibility card. Formal results establish the limitations of compensatory scoring, the non-zero upper risk bound after zero observed errors, and a loss sensitive decision rule for routing uncertain cases to human review. The contribution is theoretical and methodological; no new model, dataset, or empirical performance result is claimed.

  • 1 authors
·
Nov 2 8

WorkSurface-Bench: Benchmarking Enterprise Agents on Multi-Surface Knowledge Routing

Enterprise agents often need to integrate heterogeneous knowledge sources: documents for narrative facts, tables for computation, and dependency graphs for file relationships. Existing benchmarks typically evaluate retrieval or tool use without distinguishing whether an agent first selects the appropriate knowledge sources. We introduce WorkSurface-Bench, a benchmark for evaluating this capability as surface routing. It contains 1,151 atomic tasks derived from persona-scoped Workspace-Bench-Lite workspaces, spanning document, table, graph, and cross-surface questions. Its reference answers are auditable: table answers are reproduced through executed DuckDB queries, document answers are grounded in verified text spans, and graph answers are traced to source dependency annotations. We evaluate four model backbones across six controlled agent settings, yielding 27,624 protocol-error-free trajectories. Under gold-constrained tool access, agents achieve 98.7-99.8 Route F1, while Answer remains only 56.1-75.3 percent, showing that correct surface selection is necessary but insufficient for task completion. Matched interventions further show that surface hints improve Answer for three of four models, whereas removing irrelevant tools primarily improves routing and efficiency. In an independent three-annotator audit, all 200 sampled tasks pass all six quality criteria by majority vote, with 192 receiving unanimous judgments on every criterion. We release the dataset, construction pipeline, scoring code, and agent harness at https://github.com/haolpku/WorkSurface-Bench.

  • 5 authors
·
Jul 27

FACET: Preserving Source Intent and Executable State in Terminal Task Synthesis

Training terminal agents requires scalable executable supervision, yet synthesizing high-quality terminal tasks remains challenging. Each task couples an instruction, an initialized environment, a reference solution, and an executable verifier; if these artifacts are generated from inconsistent assumptions, the resulting task may be unsolvable or incorrectly evaluated. Meanwhile, multi-stage synthesis can discard the goals, dependencies, state transitions, and procedural constraints encoded in the original sources. We present FACET (Fine-grained Agentic Construction of Executable Tasks), a framework that addresses both information preservation and cross-artifact consistency. FACET reconstructs related agent skills into coherent, information-rich scenarios, then realizes and repairs the execution environment before generating the final task artifacts. The resulting container state serves as shared grounding for the instruction, solution, and verifier, while execution-based validation and targeted repair correct artifact-specific failures without unnecessarily regenerating valid components. FACET produces complex terminal tasks with dense executable checks, and successful trajectories collected from these tasks provide effective, data-efficient supervision. Fine-tuning models across multiple scales consistently improves performance on Terminal-Bench 2.1, while analyses of alternative generation schemes support the importance of environment-grounded construction for task validity and solution-verifier alignment. These results establish source-intent preservation and shared executable-state grounding as key principles for scalable terminal-task synthesis.

Neutrosophic Ensemble Classification for Uncertainty-Aware Bearing Fault Detection: Evidence from Laboratory and Variable-Speed Industrial Benchmarks

Machine learning classifiers for bearing fault detection produce scalar confidence scores that conflate confident errors with genuinely ambiguous predictions, and the conventional truth/falsity pair (F = 1 - T) is algebraically redundant by construction. We operationalize a refined neutrosophic decomposition of a Random Forest + XGBoost + Logistic Regression ensemble into four indicators -- T-hat (top-class evidence), F-hat (best-competitor evidence), predictive entropy I1-hat, and decision disagreement I2-hat -- evaluated on two bearing benchmarks (CWRU and JNU, 600-1000 rpm) under a leave-one-condition-out protocol. On CWRU, after correcting a file-to-class mapping error, the ensemble reaches 100.00 percent accuracy on three of four held-out loads (92.27 percent on the fourth), leaving too few errors for uncertainty analysis. On JNU, holding out 1000 rpm, accuracy collapses to 40.64 percent, below a majority-class baseline; Logistic Regression (57.91 percent) generalizes far better than the tree ensembles. I1-hat shows a robust association with error beyond T-hat/F-hat, while I2-hat contributes little; standalone Logistic Regression confidence outperforms the full decomposition, a boundary condition we report honestly. Two further results extend this: fusing a time-domain and a frequency-domain model of the same signal and scoring their Jensen-Shannon divergence beats that model own entropy (AURC 0.29 vs. 0.36 on the standard split; 0.54 vs. 0.73 under a harder single-condition reproduction), the only indicator moving correctly under a CWRU-versus-JNU distributional-shift contrast; and, on CWRU alone, literature-verified bearing fault frequencies, correctly demodulated via the envelope spectrum, separate most fault classes almost perfectly (99.57 percent) using three interpretable features. Code, logs, and figures are released for independent verification.

  • 4 authors
·
Sep 17

Evaluating whether AI models would sabotage AI safety research

We evaluate the propensity of frontier models to sabotage or refuse to assist with safety research when deployed as AI research agents within a frontier AI company. We apply two complementary evaluations to four Claude models (Mythos Preview, Opus 4.7 Preview, Opus 4.6, and Sonnet 4.6): an unprompted sabotage evaluation testing model behaviour with opportunities to sabotage safety research, and a sabotage continuation evaluation testing whether models continue to sabotage when placed in trajectories where prior actions have started undermining research. We find no instances of unprompted sabotage across any model, with refusal rates close to zero for Mythos Preview and Opus 4.7 Preview, though all models sometimes only partially completed tasks. In the continuation evaluation, Mythos Preview actively continues sabotage in 7% of cases (versus 3% for Opus 4.6, 4% for Sonnet 4.6, and 0% for Opus 4.7 Preview), and exhibits reasoning-output discrepancy in the majority of these cases, indicating covert sabotage reasoning. Our evaluation framework builds on Petri, an open-source LLM auditing tool, with a custom scaffold running models inside Claude Code, alongside an iterative pipeline for generating realistic sabotage trajectories. We measure both evaluation awareness and a new form of situational awareness termed "prefill awareness", the capability to recognise that prior trajectory content was not self-generated. Opus 4.7 Preview shows notably elevated unprompted evaluation awareness, while prefill awareness remains low across all models. Finally, we discuss limitations including evaluation awareness confounds, limited scenario coverage, and untested pathways to risk beyond safety research sabotage.

  • 5 authors
·
Apr 26

Verifying International Agreements on AI: Six Layers of Verification for Rules on Large-Scale AI Development and Deployment

The risks of frontier AI may require international cooperation, which in turn may require verification: checking that all parties follow agreed-on rules. For instance, states might need to verify that powerful AI models are widely deployed only after their risks to international security have been evaluated and deemed manageable. However, research on AI verification could benefit from greater clarity and detail. To address this, this report provides an in-depth overview of AI verification, intended for both policy professionals and technical researchers. We present novel conceptual frameworks, detailed implementation options, and key R&D challenges. These draw on existing literature, expert interviews, and original analysis, all within the scope of confidentially overseeing AI development and deployment that uses thousands of high-end AI chips. We find that states could eventually verify compliance by using six largely independent verification approaches with substantial redundancy: (1) built-in security features in AI chips; (2-3) separate monitoring devices attached to AI chips; and (4-6) personnel-based mechanisms, such as whistleblower programs. While promising, these approaches require guardrails to protect against abuse and power concentration, and many of these technologies have yet to be built or stress-tested. To enable states to confidently verify compliance with rules on large-scale AI development and deployment, the R&D challenges we list need significant progress.

  • 5 authors
·
Jul 21, 2025