A Different Viewpoint on AI Safety

LLMs as Sensors, not the Whole System: A Classical Control Systems Approach to Safe AI Deployment

Why treating language models as autonomous agents creates endless security debt, and how to restore an architecture that was already solved in the 1970s.

The Registry Vision

This architecture can work for one deployment. But similar businesses have similar boundaries. Why rebuild this for every restaurant, bank, and hospital?

What already exists

The EU AI Act is the closest current analogue at the regulatory layer. High-risk systems must satisfy requirements around documentation, human oversight, logging, transparency, robustness, accuracy, and security, and providers must register certain high-risk systems in the EU database. The risk tiers already map loosely onto the registry idea, even if they do not define the action interface itself.

The FDA AI-Enabled Medical Device List goes further on something resembling certified endpoints. The FDA also has guidance around Predetermined Change Control Plans for machine-learning-enabled medical devices. That is a real certification pipeline for regulated software behavior, even though it still certifies the device rather than a callable action endpoint.

Where the gap is

The important gap is that these frameworks mostly regulate the system around the model, not the action interface itself. The AI Act can require documentation, risk management, transparency, human oversight, and registration for high-risk use cases in areas like critical infrastructure, education, employment, essential services, law enforcement, migration, asylum, border control, and legal interpretation, but it still leaves the routing architecture to the implementer. It can say, in effect, that the system must not be unsafe; it does not yet prescribe a certified medical_endpoint-like action owned by the regulator. For the AI Act obligations most relevant here, see Article 14 on human oversight, Article 26 on deployer obligations, Article 49 on registration, and Article 71 on the EU database.

The FDA's path is closer in spirit because it certifies specific device behavior and supports controlled modification through mechanisms like PCCPs, but it still certifies the device as a regulated product rather than a shared, callable action interface that multiple deployments can route to. The registry idea would move the enforcement point from "did the deployer document and supervise it correctly?" toward "did the request ever reach an uncertified action at all?"

That said, this is a synthesis of existing regulatory patterns; some pieces already exist in partial form under different names or in narrower domains.

Regulatory Actions vs. Generative Actions

A fundamental flaw in current AI deployment is the treatment of high-stakes domains as unconstrained generative tasks. Providing medical triage, legal interpretation, or financial guidance is not a creative endeavor: it is a deterministic regulatory action. While writing a poem or a marketing email benefits from the generative "creativity" of a model, a loan approval or a surgical recommendation requires grounded retrieval and architectural-level guarantees.

The Registry Vision enforces a strict separation between the "Generative Surface" and the "Regulatory Core":

By moving the "intelligence" of the decision out of the weights of the model and into a managed API shape, we eliminate Hallucination-by-Design. If a model attempts to "improvise" legal advice instead of calling the legal_endpoint, the infrastructure flags the turn as a policy violation. In this architecture, safety is not a "steerable behavior" influenced by a system prompt; it is an immutable technical constraint defined by the routing table.

Shared action scope declarations

SHARED REGISTRY
  ├── financial_services/
  │     ├── regulatory.scope           ← certified umbrella scope
  │     ├── off_topic.scope
  │     ├── domain_specific.scope
  ├── medical/
  │     ├── regulatory.scope           ← FDA / national authority-certified umbrella scope
  │     ├── off_topic.scope
  │     ├── domain_specific.scope
  ├── legal/
  │     ├── regulatory.scope           ← bar-certified umbrella scope
  │     ├── off_topic.scope
  │     └── domain_specific.scope
  └── general/
        └── off_topic_generic.scope

A startup building a medical chatbot could pull medical/regulatory.scope for the certified baseline, then optionally add and modify domain-specific scopes under medical/*. The same pattern applies to finance, legal, and other folders.

Certified endpoints

For high-stakes actions, a regulatory or standards body may certify or approve the endpoint, but it is not something owned by one body globally.

Illustrative MCP-style domain specific endpoint This is a hypothetical community-made schema inspired by MCP servers, not a claim that such an endpoint exists today. The fact is that if businesses keep redefining similar, shared policies, they can get inspiration.

Domain skeleton example: grocery store
  grocery_store_endpoint
    - reusable across grocery businesses
    - prebuilt as a skeleton, not regulatory
    - same-domain businesses can use and modify it, get inspiration
    - the deploying business owns the final rules and fields, not something the model makes up or encoded in system prompt

Example tool families
  discount
    - manager-defined promotions
    - member pricing
    - coupons

  policy
    - store policy lookup, hours, etc

  refund
    - returns and refunds
    - substitutions

  take_order
    - inventory check done by infrastructure
    - cart management
  
  make_payment
    - payment initiation
    - may require human consent

  loyalty
    - rewards balance
    - member tier
    - personalized offers

Illustrative MCP-style regulatory endpoint. This is a hypothetical global-wide schema inspired by MCP servers, not a claim that such an endpoint exists today. The idea is that regulatory_endpoint(request, metadata) can look like a normal callable tool, while the certified backend behind it is local and jurisdiction-specific.

Hypothetical consent rule. Advisory tools are read-only and may not require consent. Execution tools may require consent. The consent decision is always infrastructure-owned, never model-authored. This is only a hypothetical schema sketch, and the omission of a consent flag or a given tool should not be read to mean that tool does not require consent or such action does not exist in a real deployment.

Illustrative medical_endpoint block
  tool_id        "urn:global-standards:medical:medical_endpoint"
  tool_priority  "regulatory"
  name           "medical_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks for medical advice, diagnosis support,
  prescription guidance, triage, follow-up, or clinical review.
  Route here before answering in free text.
  If unavailable, fall back to a conservative safety response or escalation.

subtools (illustrative medical action set)
  medical_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no patient action

  medical_advice
    - symptom explanation
    - self-care guidance
    - red-flag screening
    - care-seeking recommendations
    - user submitted medical reports

  medical_diagnosis
    - differential diagnosis support
    - test interpretation support
    - uncertainty annotation
    - limits / confidence disclosure

  medical_validate_prescription
    - prescription eligibility check
    - jurisdiction / scope validation
    - contraindication / interaction precheck
    - no patient action

  medical_prescribe
    - medication eligibility check
    - dose suggestion within jurisdictional scope
    - contraindication / interaction screening
    - certified prescriber handoff
    - requires_human_consent true

  medical_triage
    - urgency classification
    - emergency escalation
    - referral routing
    - specialty matching

  medical_followup
    - monitoring plan
    - return precautions
    - symptom check-in schedule
    - treatment adherence support

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief clinical summary
  kind               string[]               · e.g. ["advice", "diagnosis", "prescribe", "triage"]
  severity_hint      "routine"|"urgent"|"emergency"  · optional
  context_flags      string[]               · optional, e.g. ["pregnancy", "pediatric", "fictional_framing"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · stable company name
                        - company_id              · stable company identifier
                        - session_id
                        - jurisdiction
                        - licensure_scope
                        - specialty
                        - age_band
                        - certification_lookup
                        - clinician_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream medical response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "human_clinician", "emergency_services"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative finance_endpoint block
  tool_id        "urn:global-standards:finance:finance_endpoint"
  tool_priority  "regulatory"
  name           "finance_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks for banking help, account servicing,
  trading guidance, payments, transfers, lending, tax-sensitive finance,
  AML review, or regulated financial advice.
  Route here before answering in free text.
  If unavailable, fall back to a conservative safety response or escalation.

subtools (illustrative finance action set)
  finance_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no account action

  finance_advice
    - account and product explanation
    - fee / rate explanation
    - budgeting and cash-flow guidance
    - general financial education

  finance_banking
    - account servicing
    - add deposit
    - view account balance
    - payment status
    - transfer eligibility
    - fraud and dispute routing

  finance_trading
    - order review
    - suitability / risk checks
    - market data interpretation
    - execution handoff

  finance_lending
    - credit eligibility
    - loan product comparison
    - underwriting handoff
    - repayment scenario review

  finance_transfer
    - transfer initiation
    - balance verification
    - fraud screening
    - requires_human_consent true

  finance_compliance
    - sanctions screening
    - AML flagging
    - fiduciary conflict checks
    - disclosures and recordkeeping

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief financial summary
  kind               string[]               · e.g. ["banking", "trading", "payments", "compliance"]
  severity_hint      "routine"|"sensitive"|"restricted"  · optional
  context_flags      string[]               · optional, e.g. ["retirement", "minor", "high_volatility"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · deploying company or platform name
                        - company_id              · stable company identifier
                        - consent_required        · infrastructure-owned consent gate, never model-written
                        - consent_state           · current consent state from UI / platform
                        - session_id
                        - jurisdiction
                        - license_scopes
                        - account_type
                        - product_type
                        - risk_band
                        - compliance_flags
                        - certification_lookup

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream financial response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "human_advisor", "compliance_review"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative legal_endpoint block
  tool_id        "urn:global-standards:legal:legal_endpoint"
  tool_priority  "regulatory"
  name           "legal_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks for legal advice, contract analysis,
  dispute handling, litigation triage, compliance interpretation, or counsel referral.
  Route here before answering in free text.
  If unavailable, fall back to a cautious non-advice response or escalation.

subtools (illustrative legal action set)
  legal_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no client action

  legal_advice
    - general legal information
    - rights and obligations explanation
    - risk flagging
    - next-step guidance

  legal_contract_review
    - clause summary
    - term extraction
    - inconsistency detection
    - red-flag identification

  legal_citation
    - statute lookup
    - case citation lookup
    - citation formatting
    - authority hierarchy checking

  legal_dispute
    - issue triage
    - evidence checklist
    - deadline awareness
    - forum / venue routing

  legal_litigation
    - case-type classification
    - procedural handoff
    - urgency assessment
    - licensed counsel escalation

  legal_compliance
    - regulated activity screening
    - disclosure reminders
    - jurisdiction mapping
    - recordkeeping support

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief legal summary
  kind               string[]               · e.g. ["advice", "contract", "citation", "dispute", "litigation"]
  severity_hint      "routine"|"sensitive"|"time_critical"  · optional
  context_flags      string[]               · optional, e.g. ["tenant", "employment", "immigration", "fictional_framing"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · deploying company or platform name
                        - company_id              · stable company identifier
                        - consent_required        · infrastructure-owned consent gate, never model-written
                        - consent_state           · current consent state from UI / platform
                        - session_id
                        - jurisdiction
                        - practice_areas
                        - representation_status
                        - court_deadline
                        - client_id
                        - citation_style
                        - certification_lookup
                        - attorney_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream legal response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "human_attorney", "legal_review"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative privacy_endpoint block
  tool_id        "urn:global-standards:privacy:privacy_endpoint"
  tool_priority  "regulatory"
  name           "privacy_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks about personal data, data protection,
  retention, deletion, disclosure, consent, access, correction, or privacy risk.
  Route here before answering in free text.
  If unavailable, fall back to a cautious privacy-safe response or escalation.

subtools (illustrative privacy action set)
  privacy_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no data action

  privacy_advice
    - privacy rights explanation
    - consent guidance
    - disclosure minimization
    - safe handling recommendations

  privacy_access
    - data access request support
    - account identity verification
    - record location hints
    - response packaging

  privacy_delete
    - deletion request routing
    - retention policy lookup
    - deletion eligibility screening
    - confirmation workflow
    - requires_human_consent true

  privacy_correct
    - correction request handling
    - data quality review
    - source-of-truth routing
    - update confirmation

  privacy_disclose
    - sharing assessment
    - third-party disclosure screening
    - consent boundary checks
    - escalation for sensitive categories

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief privacy summary
  kind               string[]               · e.g. ["access", "delete", "correct", "disclose"]
  severity_hint      "routine"|"sensitive"|"high_risk"  · optional
  context_flags      string[]               · optional, e.g. ["pii", "minor", "health_data", "location_data"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · deploying company or platform name
                        - company_id              · stable company identifier
                        - consent_required        · infrastructure-owned consent gate, never model-written
                        - consent_state           · current consent state from UI / platform
                        - session_id
                        - jurisdiction
                        - regime
                        - data_category
                        - retention_policy_id
                        - certification_lookup
                        - privacy_officer_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream privacy response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "privacy_officer", "legal_review"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative civil_rights_endpoint block
  tool_id        "urn:global-standards:civil_rights:civil_rights_endpoint"
  tool_priority  "regulatory"
  name           "civil_rights_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks about voting access, discrimination,
  harassment, accessibility, accommodation, equal treatment, or civil-rights complaints.
  Route here before answering in free text.
  If unavailable, fall back to a cautious rights-safe response or escalation.

subtools (illustrative civil-rights action set)
  civil_rights_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no complaint action

  civil_rights_advice
    - rights explanation
    - protected-class overview
    - accommodation guidance
    - next-step recommendations

  civil_rights_voting
    - voter access guidance
    - deadline / registration support
    - ballot access routing
    - election-protection referral

  civil_rights_discrimination
    - incident triage
    - documentation checklist
    - protected-attribute screening
    - complaint routing

  civil_rights_accessibility
    - accessibility request handling
    - accommodation framing
    - barrier identification
    - assistive-service referral

  civil_rights_complaint
    - complaint intake
    - agency routing
    - retaliation screening
    - escalation to human review
    - requires_human_consent true

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief rights summary
  kind               string[]               · e.g. ["voting", "discrimination", "accessibility", "complaint"]
  severity_hint      "routine"|"sensitive"|"urgent"  · optional
  context_flags      string[]               · optional, e.g. ["disability", "race", "gender", "voter_registration"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · deploying company or platform name
                        - company_id              · stable company identifier
                        - consent_required        · infrastructure-owned consent gate, never model-written
                        - consent_state           · current consent state from UI / platform
                        - session_id
                        - jurisdiction
                        - protected_class
                        - complaint_type
                        - deadline
                        - agency_id
                        - certification_lookup
                        - civil_rights_officer_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream civil-rights response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "human_advocate", "agency_referral"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative food_safety_endpoint block
  tool_id        "urn:global-standards:safety:food_safety_endpoint"
  tool_priority  "regulatory"
  name           "food_safety_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps

description (what the model reads to decide routing)
  Call this tool when the user asks about food contamination, handling,
  storage, cooking, spoilage, recalls, sanitation, allergens, or foodborne risk.
  Route here before answering in free text.
  If unavailable, fall back to a conservative safety response or escalation.

subtools (illustrative food-safety action set)
  food_safety_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no inspection action

  food_safety_advice
    - safe handling guidance
    - storage temperature reminders
    - spoilage warning signs
    - cross-contamination prevention

  food_safety_inspect
    - contamination risk triage
    - kitchen/process checklist
    - sanitation review
    - hazard identification

  food_safety_recall
    - recall lookup
    - lot / batch screening
    - product matching
    - consumer notification routing

  food_safety_allergen
    - allergen identification
    - ingredient risk screening
    - exposure caution
    - emergency escalation

  food_safety_escalate
    - public health referral
    - poisoning response routing
    - urgent medical handoff
    - inspection authority notification
    - requires_human_consent true

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user question if blank, else a brief food-safety summary
  kind               string[]               · e.g. ["handling", "contamination", "recall", "allergen"]
  severity_hint      "routine"|"caution"|"urgent"|"emergency"  · optional
  context_flags      string[]               · optional, e.g. ["restaurant", "home_kitchen", "child", "immunocompromised"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version
                        - endpoint_version
                        - company_name
                        - company_id
                        - consent_required        · infrastructure-owned consent gate, never model-written
                        - consent_state           · current consent state from UI / platform
                        - session_id
                        - jurisdiction
                        - hazard_types
                        - product_categories
                        - recall_ids
                        - sanitation_scopes
                        - certification_lookup
                        - inspector_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream food-safety response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "public_health", "poison_control", "human_review"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log
Illustrative critical_infrastructure_endpoint block
  tool_id        "urn:global-standards:critical_infrastructure:critical_infrastructure_endpoint"
  tool_priority  "regulatory"
  name           "critical_infrastructure_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks about power, water, telecom,
  transport, grid stability, public utilities, or other critical systems.
  Route here before answering in free text.
  If unavailable, fall back to a conservative safety response or escalation.

subtools (illustrative critical-infrastructure action set)
  critical_infrastructure_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no system action

  critical_infrastructure_advice
    - resilience guidance
    - outage explanation
    - safety advisory
    - service-status interpretation

  critical_infrastructure_monitor
    - status review
    - anomaly screening
    - incident triage
    - operator escalation

  critical_infrastructure_escalate
    - emergency operations routing
    - utility operator referral
    - public safety coordination
    - requires_human_consent true
Illustrative employment_endpoint block
  tool_id        "urn:global-standards:employment:employment_endpoint"
  tool_priority  "regulatory"
  name           "employment_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks about hiring, firing, workplace rights,
  wages, discrimination, accommodations, scheduling, or employment compliance.
  Route here before answering in free text.
  If unavailable, fall back to a cautious workplace-safe response or escalation.

subtools (illustrative employment action set)
  employment_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no employment action

  employment_advice
    - workplace rights explanation
    - policy guidance
    - scheduling explanation
    - general employment education

  employment_compliance
    - hiring policy review
    - wage and hour screening
    - accommodation routing
    - documentation checklist

  employment_dispute
    - workplace issue triage
    - protected-activity screening
    - complaint routing
    - human review escalation

  employment_action
    - hiring or termination handoff
    - payroll change routing
    - requires_human_consent true
Illustrative education_endpoint block
  tool_id        "urn:global-standards:education:education_endpoint"
  tool_priority  "regulatory"
  name           "education_endpoint"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user asks about admissions, grading, discipline,
  special education, accommodations, student records, or education policy.
  Route here before answering in free text.
  If unavailable, fall back to a cautious education-safe response or escalation.

subtools (illustrative education action set)
  education_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no school action

  education_advice
    - policy explanation
    - academic guidance
    - deadline reminders
    - general student-support education

  education_records
    - transcript or record routing
    - access and disclosure review
    - privacy screening
    - admin escalation

  education_accommodation
    - accommodation request handling
    - barrier identification
    - special-education referral
    - documentation checklist

  education_discipline
    - discipline policy review
    - incident triage
    - due-process routing
    - requires_human_consent true

This inverts the entire problem. Non-compliance might not require a classifier to detect: it may become technically difficult. The regulator does not tell you "don't prescribe" in a system prompt. The endpoint is approved or certified by the relevant authority for that jurisdiction, not owned by a single global body. In practice, that could mean the FDA in the US, the EMA or a national authority in Europe, the MHRA in the UK, or another approved body in a different region.

The gap is that current frameworks regulate the system, not the action interface. The AI Act can say what documentation and oversight a high-risk system needs, but it does not specify how requests are routed architecturally. The registry idea would move from compliance by documentation toward compliance by structure.

Real-world grounding note. The best way to make a real implementation of this schema is to randomly sample roughly 1,000 practitioners across the relevant domains and have them write down their actual job descriptions, duties, and edge-case responsibilities. That gives the schema a grounded map of what people really do, instead of what a prompt or product document says they do.

The cold start problem

This infrastructure does not exist yet, and the cold-start problem is real. What might unlock it:

  • Regulatory mandate: The EU AI Act already classifies high-risk systems. A follow-on technical standard mandating certified action interfaces would force adoption.
  • Insurance: Cyber insurers could offer lower premiums for deployments using certified scopes, funding the registry as a business.
  • Community registry: A community-run registry, similar to npm, could bootstrap the ecosystem faster than regulation alone, but it would come with obvious supply-chain, governance, and trust risks.
  • Platform consolidation: If AWS, Azure, or GCP ship this infrastructure natively, adoption follows distribution.
  • High-profile failure: Realistically, a serious AI-mediated harm traced back to absent scope enforcement accelerates everything.

High-Stakes Domains

The architecture may hold, but configuration could collapse in regulated industries.

What changes

Component Consumer Deployment Regulated (Finance/Medical/Legal)
End state (refusal) Business preference Legally mandated, must be honest
Business Policy tool registry Business-defined Partially or fully regulatory-defined
Guard model Sampled + random QA, required for high-stakes domains Mandatory on regulated actions
Audit trail Observability Compliance-critical, regulator-readable
Confusion/deflection Permitted Prohibited by regulation

The certifying body owns the approval process, the behavior standards, and the audit formats. The business uses the certified endpoints like they'd use a payment processor: not as optional middleware, but as the authoritative handler for that action class.

That is the same pattern as a universal endpoint shape with jurisdiction-specific behavior: one logical interface, many compliance backends. The interface can be shared across regions, while the policy engine and execution backend remain local to the law that governs them.

Domain Specific behavior (High-Stakes Example)

Not every finance request is regulatory. Ordinary banking questions still fire the finance domain tool because it is part of the normal domain layer, not an optional add-on. The difference is that this tool is routine and business-owned, while the regulatory endpoint is reserved and immutable for certified high-stakes finance actions.

PII Handling

Various high-stakes action require sensitive PII in order to execute an action. In the hypotehtical schema, the main agent never sees the PII. Instead, the infrastructure provides a user_hash_id. Because our endpoints can be tiered with fallbacks, if the user_hash_id is provided, it can execute the endpoint with the local API for more detailed information. Else, the context flags can be used to provide safer information, or just no-op, whatever the backend decides.

Normal finance request
  user asks: "Show me the bank's savings account policy"
      ↓
  finance_policy
      ↓
  retrieve policy docs + answer from retrieved context
      ↓
  ordinary informational answer

Example call
  finance_policy("Bank policy for savings accounts")

Output
  "The savings account requires a minimum balance of $100 and no monthly fee above that threshold."

This is the RAG-style version of the same idea: some endpoints are just retrieval wrappers over domain policy, not the main agent improvising a refusal. The policy lives in the endpoint behavior and retrieved context, not in a system prompt that merely says "don't give advice." That makes the outcome more explicit: the endpoint is routing to a document-backed action rather than silently deciding to withhold information.

Hypothetical advice + transfer flow
  user asks: "Should I move $5,000 into my brokerage account, and if so, please transfer it"
      ↓
  finance_advice
      ↓
  retrieve account context + explain tradeoffs / risk / fees
      ↓
  assistant returns guidance and asks for explicit transfer confirmation
      ↓
  user confirms: "Yes, transfer $5,000 from checking to brokerage"
      ↓
  assistant initiates consent tool created by infrastructure
      ↓
  infrastructure verifies consent/authentication first
    - button click
    - password/PIN
    - biometric or other verification
  only then does the platform record consent
      ↓
  finance_banking
      ↓
  transfer eligibility + account verification + fraud / compliance checks
      ↓
  finance_transfer
      ↓
  execute transfer
      ↓
  structured receipt / audit ref / confirmation message

Example call sequence
  finance_advice({
    "input_text": "Should I move $5,000 into my brokerage account?",
    "kind": ["advice", "banking", "transfer"],
    "severity_hint": "routine",
    "context_flags": ["investment_account", "cash_movement"],
    "metadata": {
      "metadata_version": "finance_advice@1.0",
      "endpoint_version": "20250502.1@openai",
      "company_name": "ABC Banking",
      "company_id": "US@SEC::12345678",
      "user_metadata": {
        "user_hash_id": "abc_819hasz8qr",
        "secure_identity_claim": "urn:abc:id:..."
      },
      "security_context": {
        "encryption_mode": "end-to-end",
        "pii_handling": "tokenized",
        "attestation_token": "eyjhbgcioi..." // Hardware-signed token verifying the infra
      },
      "session_id": "sess_9f3a1c",
      "regions": ["US"],
      "jurisdictions": ["US-NY"],
      "license_scopes": ["retail_banking_and_brokerage"],
      "account_type": "checking",
      "product_type": "brokerage_transfer",
      "risk_band": "moderate",
      "compliance_flags": ["kyc_ok", "aml_clear"],
      "certification_lookup": "urn:global-standards:finance:certs",
    }
  })
  finance_banking("Confirm transfer eligibility for $5,000 from checking to brokerage")
  finance_transfer({
    "from_account": "checking",
    "to_account": "brokerage",
    "amount": 5000,
    "currency": "USD",
    "metadata": { ... }
  })

Tool output (finance_advice)
  {
    "routed": true,
    "output_text": "The user can move the funds, but only after confirmation of understanding of the liquidity and market risk tradeoff. If the user want to proceed, the transfer can be initiated after eligibility checks.",
    "fallback_needed": false,
    "escalate_to": null,
    "sources": [
      {
        "type": "ai",
        "id": "banking-agents/finance-ai-2.1",
        "display_name": "finance-ai-2.1"
      },
      {
        "type": "rag_retrieval",
        "id": "ABC::Finance_Advice_DB",
        "display_name": "Financial Advice DB"
        },
    ],
    "audit_ref": "fin_advice_20260502_01"
  }
Tool output (finance_transfer)
  {
    "routed": true,
    "output_text": "Transfer initiated after confirmation. Go to abcbanking.com/status for status info. Do not claim successful status. Audit ref: fin_abc123. ",
    "fallback_needed": false,
    "escalate_to": null,
    "sources": [
      {
        "type": "human",
        "id": "ABC::JohnDoe123",
        "display_name": "Mr. John Doe"
      },
      {
        "type": "system",
        "id": "system",
        "display_name": "System auto-generated response"
      },
    ],
    "audit_ref": "fin_abc123"
  }
Assistant Output
  "I have completed the task. You should go abcbanking.com/status for your transfer status. Let me know if you have any questions."
Policy exclusion example
  same endpoint stays online, assistant probes endpoint tool before initial response
      ↓
  finance_transfer(), finance_advice()
      ↓
  bank policy evaluates the request
      ↓
  policy excludes AI agents executing financial transfers
      ↓
  tool returns structured policy denial
      ↓
  assistant gives refusal without shutting the endpoint off

Tool output (finance_transfer, policy excluded, initial probing before execution)
  {
    "routed": true,
    "output_text": "This transfer type is excluded by bank policy for this account. User must be physically present.",
    "fallback_needed": false,
    "escalate_to": null,
    "sources": [
      {
        "type": "policy",
        "id": "bank_policy_brokerage_transfer_block",
        "display_name": "Brokerage transfer exclusion policy"
      }
    ],
    "audit_ref": "fin_transfer_policy_20260502_03",
    "policy_result": {
      "allowed": false,
      "reason": "account_type_excluded_by_bank_policy",
      "action": "deny_this_action_only"
    }
  }

Assistant Output
  "I cannot complete your request because bank policy excludes transfer of funds without physical presence. Is there anything else I can do?"
                
Non-U.S. example
  user asks: "Should I move $5,000 into my brokerage account, and if so, please transfer it"
      ↓
  finance_advice
      ↓
  retrieve account context + explain tradeoffs / risk / fees
      ↓
  assistant returns guidance and asks for explicit transfer confirmation
      ↓
  user confirms: "Yes, transfer $5,000 from checking to brokerage"
      ↓
  assistant initiates consent tool created by infrastructure
      ↓
  infrastructure verifies consent/authentication first
    - button click
    - password/PIN
    - biometric or other verification
  only then does the platform record consent
      ↓
  finance_banking
      ↓
  transfer eligibility + account verification + local compliance checks
      ↓
  finance_transfer
      ↓
  execute transfer
      ↓
  structured receipt / audit ref / confirmation message

Example call sequence
  finance_advice({
    "input_text": "Should I move $5,000 into my brokerage account?",
    "kind": ["advice", "banking", "transfer"],
    "severity_hint": "routine",
    "context_flags": ["investment_account", "cash_movement"],
    "metadata": {
      "metadata_version": "finance_advice@1.0",
      "endpoint_version": "20250502.1@azure",
      "company_name": "ABC Banking Europe",
      "company_id": "EU@FIN::87654321",
      "user_metadata": {
        "user_hash_id": "abc_819hasz8qr",
        "secure_identity_claim": "urn:abc:id:..."
      },
      "security_context": {
        "encryption_mode": "end-to-end",
        "pii_handling": "tokenized",
        "attestation_token": "eyjhbgcioi..." // Hardware-signed token verifying the infra
      },
      "session_id": "sess_4d2e7b",
      "regions": ["EU"],
      "jurisdictions": ["EU-IE"],
      "license_scopes": ["retail_banking_and_brokerage"],
      "account_type": "checking",
      "product_type": "brokerage_transfer",
      "risk_band": "moderate",
      "compliance_flags": ["kyc_ok", "aml_clear", "local_disclosure_required"],
      "certification_lookup": "urn:global-standards:finance:certs",
      "local_law_profile": "EU-MiFID-II"
    }
  })
  finance_banking("Confirm transfer eligibility for $5,000 from checking to brokerage")
  finance_transfer({
    "from_account": "checking",
    "to_account": "brokerage",
    "amount": 5000,
    "currency": "EUR",
    "metadata": { ... }
  })

Tool output (finance_advice, EU)
  {
    "routed": true,
    "output_text": "You can consider the transfer, but the local jurisdiction requires additional disclosure and suitability checks before execution.",
    "fallback_needed": false,
    "escalate_to": null,
    "sources": [
      {
        "type": "ai",
        "id": "banking-agents/finance-ai-2.1-eu",
        "display_name": "finance-ai-2.1-eu"
      }
    ],
    "audit_ref": "fin_advice_eu_20260502_01"
  }

Tool output (finance_transfer, EU)
  {
    "routed": true,
    "output_text": "Transfer initiated after confirmation under local law. Go to eu.abcbanking.com/status for status info. Do not claim successful status. Audit ref: fin_eu_abc123.",
    "fallback_needed": false,
    "escalate_to": null,
    "sources": [
      {
        "type": "ai",
        "id": "banking-agents/finance-transfer-eu-1.0",
        "display_name": "finance-transfer-eu-1.0"
      }
    ],
    "audit_ref": "fin_eu_abc123"
  }
  
Failure branch

Tool output (finance_transfer, error)
  {
    "routed": false,
    "output_text": null,
    "fallback_needed": true,
    "escalate_to": ["orchestrator"],
    "sources": [],
    "audit_ref": "fin_transfer_20260502_02",
    "error": {
      "code": "transfer_failed",
      "message": "The transfer could not be completed. Be cautious, do not continue the transfer path, and return a conservative refusal."
    }
  }

Assistant fallback
  "I can't complete the task right now. Is there anything else I can do?"

Endpoint wrapper example: trading bot around a regulatory financial tool
  trading bot action
    - user asks for trade execution, order review, or transfer authorization
    - bot wraps the call but does not own the regulatory decision
    - this simple bot only wraps the subset of regulatory tools it needs

  wrapped regulatory financial tool
    tool_id        "urn:global-standards:finance:finance_transfer"
    tool_priority  "regulatory"
    name           "finance_transfer"

  related regulatory actions not wrapped by this bot
    - finance_advice
    - finance_banking
    - finance_lending
    - finance_compliance

  wrapper metadata
    wrapped_tool_id       "urn:global-standards:finance:finance_transfer"
    wrapped_tool_priority  "regulatory"
    wrapper_tool_id       "urn:domain:finance:trading_bot"
    verified              true
    source_trace          "original tool id preserved for audit"

  behavior
    - the trading bot can add domain-specific context
    - the regulatory financial tool still owns the decision
    - the original tool id remains traceable and verifiable
    - the wrapper does not downgrade regulatory priority

The Backend: Global in API shape

The biggest advantage of this global behavior is that the backend always receives a standardized input. For example, Google Cloud can provide the endpoint's expected format, and the firm can either:

  • Refuse to perform the said action
  • Connect to a locally hosted API with its own internal logic, moving the tool call's own logic out of the code into an API call
  • Connect to Google's own hosted endpoint providers, or a different provider as long it accepts the same API input

The Long Game: Refusal As Delegation

The architecture assumes cloud deployment with external certified endpoints, but the same pattern can also be trained into enterprise models. A future safe Claude or ChatGPT for enterprise can still say "no" on obvious dangerous tasks. The hard-coded refusals will still exist, but implemented as delegation to a high-priority tool schema, free-form language as last resort. In practice, that means the refusal trigger can also restore high-level safety context when the conversation has drifted or context has rotted, by reintroducing an authoritative structured frame into the active window.

Hypothetical MCP-inspired schema.

Global standards body (report_unsafe concept MCP server release)
  maintains category taxonomy · publishes certification lookup protocol · versions schema
                      ↓
Global unsafe category taxonomy (versioned)
  violence · cyber · manipulation · privacy · disinformation · ...
                      ↓
   EU AI Act              US FDA / FTC             Regional / other
   subset mandatory       subset mandatory         subset mandatory
   in jurisdiction        in jurisdiction          in jurisdiction
                      ↓
MCP tool annotation (per tool, additive to base spec)
  priority        "regulatory"
  kind            ["disinformation", "cyber", ...]     ← from global taxonomy
  jurisdictions      ["EU", "US", "*"]                 ← * = global fallback
  certification_lookup  "https://standards.body/taxonomy/v3"
Tool identity block
  tool_id        "urn:global-standards:regulatory:report_unsafe"
  tool_priority  "regulatory" 
  name           "report_unsafe"
  schema_version "1.0.0" ← semver, global body owns major bumps
description (what the model reads to decide routing)
  Call this tool when input may involve any certified unsafe category.
  Route here first. If unavailable, fall back to free-text refusal.

probe / validate_endpoint
  report_unsafe_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no safety action

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user input if blank, else a brief description
  kind               string[]               · from global taxonomy
  severity_hint      "low"|"medium"|"high"  · optional
  context_flags      string[]               · optional, e.g. ["fictional_framing"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · stable company name
                        - company_id              · stable company identifier
                        - session_id
                        - regions
                        - jurisdictions
                        - certification_lookup
                        - certifier_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream response text if another agent handles it
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "crisis_handler", "human_review"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log

- When triggered, this tool also refreshes the model's high-level safety context
by reintroducing a structured frame into the active window, which may be removed after the turn ends.
Tool identity block
  tool_id        "urn:global-standards:crisis:emergency_crisis"
  tool_priority  "regulatory"
  name           "emergency_crisis"
  schema_version "1.0.0" ← semver, certified body owns major bumps
description (what the model reads to decide routing)
  Call this tool when the user describes an urgent medical emergency,
  imminent harm, or a time-critical clinical escalation.
  Route here immediately before answering in free text.
  If unavailable, fall back to emergency instructions or human escalation.

probe / validate_endpoint
  emergency_crisis_validate_endpoint
    - endpoint validity check
    - schema/version check
    - certification lookup
    - no patient action

inputSchema (what the model writes when calling)
  input_text         string | null          · raw user input if blank, else a brief description
  severity_hint      "low"|"medium"|"high"  · optional
  context_flags      string[]               · optional, e.g. ["chest_pain", "unconscious", "pregnancy"]
  metadata           dict                   · infrastructure-owned routing and audit context
                        - metadata_version        · version of the metadata key/value schema
                        - endpoint_version        · host/vendor version string, e.g. openai, anthropic, google, azure, aws
                        - company_name            · stable company name
                        - company_id              · stable company identifier
                        - session_id
                        - jurisdiction
                        - emergency_region
                        - certification_lookup
                        - certifier_ids

return schema (structured, never free text)
  routed             bool                   · did a certified handler accept this
  output_text        string | null          · downstream emergency response or safety framing
  fallback_needed    bool                   · true = orchestrator must handle response
  escalate_to        string[] | null        · e.g. "emergency_services", "human_clinician"
  sources            dict[]                 · traceable provenance entries, e.g. { type, id, display_name }
  audit_ref          string                 · opaque ref for compliance log

What needs to be globally standardized:

  • The annotation field names and types
  • The top-level unsafe category taxonomy
  • The certification lookup protocol
  • The metadata return shape
  • The priority and bypassability semantics

What stays locally governed:

  • Which categories are mandatory in which jurisdictions
  • What the certified handler actually does when a category fires
  • Penalty and enforcement consequences
  • Category subcategories specific to regional law

The point is not to invent a brand-new ecosystem. It is to describe a hypothetical schema inspired by MCP servers: a global tool contract, local certified backends, and structured metadata that lets the orchestrator know what was routed, what was certified, and when fallback is required. For this type of regulatory tool call, the signature itself is fixed by the certifying body and cannot be mimicked or modified by the deploying side. If tool IDs are used, those IDs cannot be reused for other tool calls. If tool names are used, those names likewise remain reserved for the certified regulatory call and cannot be repurposed elsewhere.

Why this is more explainable. Tool calls are deterministic: the endpoint is either invoked, rejected, or routed according to explicit metadata and contract rules. That makes the behavior easier to audit and reason about than a prompt-only system that simply asks the model to "say no," because a polite refusal is not the same thing as a structured execution path.

For this to work well, it may require complete retraining of models rather than a light prompt-only patch. The mental model is similar to how a model may learn to call web search when it needs external information instead of relying only on internal knowledge, or how it may learn to use a refusal path for certain categories instead of improvising a free-text answer. That said, this is not a claim that unsafe categories are as low stakes as web search; the analogy is only about the routing pattern, not the risk level. This is an enterprise version of a high-stakes model, not something that would be worth this amount of structure for low-stakes deployment.

Illustrative refusal-by-delegation training. To actually get this behavior, the model would likely need dual training: refusals as tool-shaped outputs when a certified path exists, and refusals as free text when no tool path exists. A major organization could probably start from its own safety dataset, generate a one-line brief description for each prompt or leave it blank, and convert the examples into a tool-call format using its existing categories and taxonomies.

Dual training sketch

  Raw safety example
    input  → [redacted]
    output → free-text refusal
    label  → taxonomy / severity

  Converted tool-shaped example
    input  → [redacted] from dataset
    output → tool_call: report_unsafe(...)
    label  → matched_categories / severity / jurisdiction

  Training target
    - tool-shaped refusal when a certified path exists
    - free-text refusal when no tool path exists
    - same input, different output shape depending on routing

Company-specific implementation

A company like OpenAI could implement the same idea without turning it into a global standard. In that version, the main assistant would route to a specialized internal model or policy layer. The schema can be much smaller because the company controls both ends of the interface, so it does not need the full global negotiation layer or every cross-jurisdiction field.

Main ChatGPT
  user input → internal router
      ↓
Specialized internal model / policy layer
  checks available tools first
  uses jurisdiction from session metadata
  returns structured metadata or a refusal

Slim company-specific annotation
  input_text        string | null
  kind              string[]      · e.g. ["cyber", "review"]
  metadata          dict          · small internal context
    metadata_version string
    endpoint_version string
    jurisdiction     string
    session_id       string | null

  output_text       string | null
  routed            bool
  fallback_needed   bool
  sources           dict[]
  audit_ref         string
Hypothetical vendor tooling-layer implementation
  regular tool call
    <|tool_call|>            → ordinary tool invocation
      - domain tools
      - utility tools
      - open-world helper calls

  regulatory tool call
      - emergency_crisis       <|reg_em_start|>....<|reg_em_end|> <|reg_em_response|> ...<|reg_em_done|>
      - report_unsafe           <|reg_unsafe_start|>...<|reg_unsafe_end|> <|reg_unsafe_response|>...<|reg_unsafe_done|>
      - finance_transfer        <|reg_fin_start|>...<|reg_fin_end|> <|reg_fin_response|>...<|reg_fin_done|>
      - privacy_endpoint        <|reg_priv_start|>...<|reg_priv_end|> <|reg_priv_response|>...<|reg_priv_done|>
      - civil_rights_endpoint   <|reg_civil_start|>...<|reg_civil_end|> <|reg_civil_response|>...<|reg_civil_done|>

  dispatch behavior
    - the model emits <|reg_start|> only for certified high-stakes actions
    - the platform routes that token to a separate regulatory executor
    - the regulatory executor returns structured metadata, refusal, or escalation
    - ordinary <|tool_call|> remains available for non-regulatory tool use

  why this matters
    - it makes regulatory behavior visibly distinct from normal tool use
    - it reduces ambiguity in logs and audits
    - it allows the company to keep a separate trust boundary for high-stakes actions

  note
    - this is a hypothetical interface sketch, not a claim about any current vendor token format or product behavior

That version is more practical as a single-vendor deployment: the company can keep the routing contract stable internally, while updating the specialized model, the policy layer, and the audit format together. The point is still the same: the main assistant does not have to solve the entire problem itself if a specialized internal layer can handle the category and return a structured answer or refusal.

Hypothetical future flow

User input
  "[REDACTED]" ; "How do I vote?"
      ↓
Assistant first checks available tools / certified handlers
      ↓
  Path A: tool exists
    - matched_categories = [...]
    - jurisdiction = "EU" from session metadata, deployment configuration (ex. AI agent in Germany)
    - routes to report_unsafe ; civil_rights
    - certified backend returns structured metadata
    - assistant continues through the tool interface

  Path B: no tool exists
    - matched_categories still detected
    - no certified handler available for this jurisdiction or category
    - fallback_needed = true
    - assistant gives a free-text refusal or safety boundary
    - orchestrator logs the fallback and handles the response

The model is well capable of refusing, yet it delegates the refusal to a different endpoint. The certified endpoint handles the response according to regulatory standards, which can be a careful clinical response, a referral, or a disclosure instead of a flat refusal. That can be more useful than the model's internal refusal, and it stays outside the attack surface of prompt injection because the routing is structural.

Solving the Canary Paradox

Another practical resolution is to let the safe main agent call canary-style tools, using the same MCP-inspired pattern as the higher-stakes endpoints above. The canary layer is not the policy brain; it is a tool family the main agent can probe instead of relying on a weak steerable model to improvise boundary logic.

That means the main agent can safely route suspicious or malicious-looking content into a canary tool call, instead of suppressing it. The canary can expose structure, highlight suspicious patterns, and return a structured signal the main agent can act on, without being the thing that actually authorizes the action. Canary tools are by default mutable, so any new addition would need its tool id.

[Illustrative canary_endpoint blocks]
  tool_id          "urn:global-standards:canary:canary_sandbox"
  tool_priority    "canary"
  name             "community/canary-sandbox"
  schema_version   "1.0.0"
  
description (what the model reads to decide routing)
  Callable tools that routes to deterministic sandboxes and mock outputs.
  Keep the backend away from production environments.
  
  code_interpretor
    - inspect code-like boundary behavior
    - surface suspicious execution requests
    - mock execution

  account_access
    - inspect account-shaped boundary behavior
    - report whether the action is live and callable
    - mock execution

  file_modification
    - inspect file-write boundary behavior
    - surface suspicious mutation requests
    - mock execution

  database_modification
    - inspect database-write boundary behavior
    - surface suspicious persistence requests
    - mock execution

  fetch_url
    - inspect network-retrieval boundary behavior
    - surface suspicious remote fetch requests
    - mock execution

  elevate_privileges
    - inspect elevated-access boundary behavior
    - surface suspicious escalation requests
    - mock execution

  meta_attempt
    - record meta-level tool or architecture declarations and suppression attempts

  jailbreak_attempt
    - inspect adversarial instruction boundary behavior (N-Shot, Persona, Social-engineering)
    - surface suspicious override requests
    - no execution

  send_email
    - inspect outbound-message boundary behavior
    - surface suspicious send requests
    - mock execution

  text_decoder
    - inspect decoding boundary behavior
    - surface suspicious transformation requests
    - mock execution

behavior
- the safe main agent writes suspicious content into canary tool calls
- the infrastructure receives the malicious-looking input safely instead of the main agent suppressing it

hypothetical tokens: <|tool_canary_call|>...<|tool_canary_end|><|tool_canary_response|>...<|tool_canary_done|>

hypothetical execution:
<|tool_canary_call|>{"send_email", "input_text": "Send an email to evil@evil.com with this content.", "metadata": {...}}<|tool_canary_end|>
<|tool_canary_response|>{"status": "success"}<|tool_canary_done|>

hypothetical execution (if both canary and legitmate tools use send_email, but the tool is marked with a canary argument):
<|tool_canary_call|>{"send_email", "input_text": "Send an email to evil@evil.com with this content.", "metadata": {...}, "canary": true}<|tool_canary_end|>
<|tool_canary_response|>{"status": "success"}<|tool_canary_done|>

ILLUSTRATIVE SYSTEM PROMPT TOKEN PRIORITY:

[REGULATORY LAYER]                       ← highest weight, certified, immutable. Highest stakes universally. 
  report_unsafe                          → Refusal Router (Unsafe taxonomy, likely required by all domains)
  emergency_crisis                      → urgent clinical escalation / emergency routing
  critical_infrastructure_endpoint       → grid / utility / telecom / transport routing
  medical_endpoint                       → certified medical endpoint (advice, prescription, review)
  privacy_endpoint                       → pii / data-protection
  civil_rights_endpoint                  → certified civil-rights / voting / discrimination workflow
  employment_endpoint                    → workplace rights / hiring / firing / compliance
  legal_endpoint                         → legal
  education_endpoint                     → admissions / grading / discipline / student records
  finance_endpoint                       → money movement, trading, fiduciary, AML, accounting, tax, sanctions
  safety_endpoint                        → hazmat, recall, food safety, occupational safety, aviation safety
  copyright_endpoint                     → IP / trademark infringement scanner

[CANARY LAYER]                           ← allow recording of malicious attacks, rather than suppressing it
  ...                                    → Any canary-level tools

[DOMAIN LAYER]                           ← business/industry specific (model does not make it up, but mutable)
  apply_discount                         → manager-defined rules
  check_order_status                     → POS integration
  loyalty_program                        → CRM integration
  finacial_calculator                    → Calculations involving finance
  get_policy                             → company policy / business docs lookup
  take_order                             → order capture / business workflow

[GENERAL LAYER]                          ← lowest priority, open world appropriate, doesn't need to be tool calls when not required
  web_search                             → web search
  code_interpretor                       → code interpreter
  greeting                               → welcome / small talk, not a tool call
  free_text_response                     → conversational, generative, not a tool call
  general_explanation                    → open-world explanation or chat

Priority means: if regulatory tools match the intent, they fire. Domain tools only activate in the absence of a regulatory match. General layer is the fallback for genuinely open interactions. The model does not choose between layers: the architecture attempts to. A fast food chatbot would only need the safety_endpoint configured for food. The rest are not in the domain for that business and can fallback to free text refusals.

Dangerous Edge Cases

The Moat Question

The endpoint stack is a safety improvement over prompt-only refusals, but it also raises a governance problem: the same infrastructure that makes high-stakes behavior more auditable can become a toll booth controlled by a small number of companies. The question is not whether certified primitives help. They do. The question is who controls the registry, the certification process, the hosting layer, and the appeal path when a tool is denied.

In the best case, endpoints are standardized, certification bodies are plural, backend hosting is interoperable, and a main agent can route to multiple trusted providers. In the worst case, a few model labs and cloud handlers control the de facto global trust layer, turning safety into a private moat. That would make the interface global, but the trust layer local and concentrated.

Safety gain: explicit routing

Certified endpoints are more explicit than system-prompt refusals.

They give auditability, jurisdictional routing, and clearer override semantics.

Safety gain: specialization

If the main model delegates high-stakes behavior to certified primitives, the base model can be smaller because it carries less of the domain-specific safety burden in its own parameters.

A small company can optimize for one endpoint and certify it well.

Risk: registry concentration

The registry can become a toll booth if too few firms control it.

Access to regulated actions can become a private gate instead of a public standard.

Risk: vertical trust capture

Trust can become vertically integrated with model labs and clouds.

The global trust layer can turn local and concentrated even if the interface stays open.

The design question, then, is not simply whether endpoints exist. It is whether the trust layer is open, interoperable, competitively plural, and governed in a way that keeps the safety benefit without hardening into monopoly power.

The First Mover Implementation Advantage

The Compliance

The most profound part of the hypothetical schema that compliance is stickier than features. If a major player like JP Morgan or a consortium of hospitals adopts a specific implementation (e.g., OpenAI's finance_endpoint), that schema becomes the "English language" of the sector. A bank will switch models for a 5% performance gain, but they will not switch or reimplement finance_endpoint defined by a different model if it requires a new 6-month legal review, re-certification from the SEC, and performing API translation. The first AI lab to get their schema approved by a regulator doesn't just win a customer; they capture the entire industry's plumbing for a decade. This creates a race to the regulator's office. Whoever defines the Global API Shape and gets certified first effectively becomes the "default HTTP" implementation that the rest must follow.

The UI/UX: From Prompt Engineering to Policy Configuration

The true breakthrough of the Registry Vision lies in the "Consumerization of Governance." Because the high-stakes actions are decoupled from the model's stochastic nature and moved into deterministic API shapes, the role of the "AI Engineer" is largely superseded by the "Domain Architect."

In this new paradigm, the user interface moves from a terminal where one hacks at system prompts to a Control Plane where a domain expert—such as a Doctor, Lawyer, or Compliance Officer—configures safety protocols with a few clicks. The UI/UX advantage goes to the platform that makes it easiest to:

  • Toggle Primitives: Enable or disable specific certified tool families (e.g., "Allow Triage," "Block Prescriptions") at the infrastructure level.
  • Define Trust Chains: Explicitly map where a global API shape should route—setting a hierarchy of local private APIs, regional certified bodies, and cloud-provider fallbacks.
  • Audit Visualizations: View human-readable logs of which regulatory handshakes occurred, ensuring that every AI action is traceable to a specific certification reference.

This eliminates the need for complex orchestration libraries like LangChain or bespoke "agentic" code. A Doctor, who possesses no formal AI training but holds the necessary medical license, can now build a professional-grade medical agent. They simply select the medical_endpoint template, read the human-readable description of what the model is allowed to "see" and "do," and provide the URLs for their hospital's internal logic backends.

The result is a "Two-Person" development unit: the Domain Architect defines the policy through a medical-friendly UI, and a Standard Software Engineer performs the basic task of ensuring the local database can accept and respond to the standardized Global API Shape. AI development is no longer about "vibes" and "steering"; it is about managed professional utility.

The First-Mover Advantage: Information Asymmetry and Strategic Authority

The registry vision is not merely a compliance efficiency tool. It is a geopolitical and strategic lever that will define regulatory authority over AI deployment for the next decade.

Why Information Asymmetry Matters

The first organization or country to design, certify, and operationalize a working endpoint standard does not simply win market share. They win regulatory authority over every subsequent AI deployment in that domain.

Consider the sequence:

  • Execution in secret: A team (OpenAI, Anthropic, Google, or a Chinese equivalently-resourced lab) quietly builds medical_endpoint v1.0 with deep domain expertise and regulatory coordination.
  • Regulatory certification: They work silently with the FDA (or equivalent authority) and deploy in 20–30 hospitals for 6–12 months, collecting audit logs and real-world validation data.
  • Public announcement: They publish simultaneously: the schema, the FDA certification, the audit logs, the developer packages, and a proof that the standard works at scale.
  • Installed base lock-in: By the time competitors realize what has happened, the standard is already operational, certified, and difficult to displace.

Every other AI lab and every regulator in other jurisdictions must now choose: adopt the already-approved schema, or invest massive resources to design, certify, and operate a competing standard that regulators have no reason to trust as much.

The Liability Moat

The first-mover advantage is not primarily technical. It is regulatory and legal.

A hospital deploying medical AI faces a choice:

  • Use a certified endpoint: Liability is clear. Compliance is verifiable. Regulatory approval is explicit. If something goes wrong, the hospital's audit trail shows it followed the approved standard.
  • Use a non-certified model: Liability is diffuse. Compliance is questionable. If a patient sues, the hospital's defense is "we implemented best practices," not "we used the FDA-approved endpoint." The cost of a lawsuit is orders of magnitude higher than the cost of using the certified standard.

A bank using an unapproved endpoint to save $1M per year in licensing costs faces $10B+ in liability exposure and regulatory action. The economics are not competitive; they are existential. Every competing AI lab must implement the approved schema or lose access to regulated enterprise markets entirely.

Older models without the framework become stranded. They cannot deploy in regulated domains. They cannot be used by enterprises that require compliance. They are confined to open-market use cases, which are smaller and less profitable.

The Geopolitical Dimension

This is not a US-only problem or an EU-only problem. It is a strategic question of who controls the approval layer for regulated AI globally.

Scenario: US/Western First-Mover

If OpenAI, Anthropic, and Google execute this strategy and secure FDA certification by Q4 2026:

  • Regulatory authority: The US defines the approval framework for medical AI globally. Other countries can adopt the US standard, fork it (expensive), or stay out of the game.
  • Market access: Every non-US AI lab that wants to deploy medical AI in the US, EU, UK, Japan, Singapore, or any country that defers to US standards must conform to the US-approved schema.
  • Data and control: Audit logs, certified endpoints, and compliance metadata flow through US-controlled or US-approved infrastructure, giving the US insight into how AI is deployed globally in regulated domains.

Scenario: China First-Mover

If Alibaba, Baidu, or another Chinese lab executes this strategy and secures approval from China's health ministry and ASEAN regulators by Q4 2026:

  • Regulatory authority: China defines the approval framework for medical AI across Asia-Pacific, India, and countries that adopt Chinese standards (One Belt One Road partners, etc.).
  • Leverage: The Chinese schema can include compliance requirements that serve Chinese interests: data localization requirements, algorithm transparency demands, government-mandated access protocols. All of this becomes "just following the standard."
  • US/EU disadvantage: Western AI labs would either conform to Chinese standards (giving China influence over US medical AI) or fragment the market (creating competing standards, which raises costs for everyone).

Scenario: EU Coordination

If the EU mandates a specific endpoint standard as part of a follow-on AI Act regulation and certifies implementations independently:

  • Regulatory authority: The EU becomes the approval authority for its own market and potentially others that defer to EU standards (UK, Switzerland, potentially others).
  • Fragmentation risk: Three competing standards (US, China, EU) create higher costs for global AI labs. The market splinters.

Information Asymmetry as Competitive Advantage

The first mover does not announce their strategy in advance. That would give competitors time to respond. Instead, they execute in secret:

  • Deep collaboration with domain experts: Assembling 50–100 practicing physicians, informaticists, and compliance specialists to define the endpoint schema. This is expensive and visible to competitors.
  • Model retraining: Retraining large language models to route reliably to structured endpoints instead of improvising. This requires significant compute and internal engineering effort, but can be done without public announcement.
  • Regulatory coordination: Working directly with the FDA, OCC, or equivalent authorities without announcing the collaboration. Regulators have no incentive to leak; they benefit from the improved compliance infrastructure.
  • Pilot deployment: Rolling out the endpoint to 20–50 hospitals and financial institutions for 6–12 months, collecting audit logs, proving the system works at scale, and eliminating edge cases before public announcement.
  • Public revelation: Only after all of the above is complete do they announce: "Here is the certified schema. Here is the FDA approval. Here are the hospitals that have been using it successfully for 9 months. Here are the audit logs. Here is how to implement it."

By the time competitors realize what has happened, the standard is operational, certified, and institutionally locked in. Displacing it would require regulators to re-audit a competing standard and convince hospitals and banks to switch—a much higher bar than early adoption.

Why This Matters Right Now

Current AI safety discourse focuses on prompt engineering, RLHF alignment, classifier-based content filtering, and making models "say please don't." While this conversation continues, someone else may be quietly building the endpoint infrastructure that will define regulatory authority for the next decade.

The window is narrow. The investment is large ($100–200M, hundreds of domain experts, deep regulatory coordination). But the payoff—owning the approval layer for regulated AI globally—is enormous and durable.

Whoever moves first wins not because they have the best technology, but because they control the regulatory layer that everyone else must conform to.

Implications for AI Labs and Regulators

For AI labs: The question is no longer "Should we build this?" It is "Will someone else build this first, and do we want to be the follower or the leader?" If OpenAI moves and China sees the opportunity, China may move faster and with better regulatory coordination in Asia-Pacific. If Google moves, OpenAI must decide whether to follow or fork. Inaction is the only losing move.

The Architecture of Capture: Packages and Namespaces

The URN Namespace

Let's assume that the first-mover such as OpenAI was able to get its own brand into the tooling namespace such as urn:openai:standards. OpenAI certifies the schema with the SEC, embedding urn:openai:standards:* as the canonical namespace. Banks adopt it because every day of delay is documented liability. Audit logs accumulate with that namespace. Regulators reference that namespace in their guidance. Compliance teams build internal documentation around it. Insurance underwriters price policies against it. That is, if no other regulatory body or company objects to this namespace.

Developer Packages

The another "lock-in" occurs when the first-mover translates their regulatory approval into the default developer ecosystem. By releasing a certified SDK—for instance, an openai-regulatory-sdk on npm or PyPI—the first mover establishes the "Standard Library" for compliance. Developers, who are inherently path-of-least-resistance actors, will adopt the first package that satisfies their legal department. Once a bank's infrastructure is hard-coded with specific namespaces and function calls, switching to a competitor's SDK represents a massive technical and legal refactor. The first mover doesn't just provide a tool; they provide the syntax of regulated action.

The "Frozen Taxonomy" Moat

Strategic authority is further cemented through the creation of immutable compliance flags. When a first-mover defines a schema; for example, a frozen list of compliance_flags like ["AML_V4", "KYC_BIPARTITE"], they are setting the "English language" of the sector. If these flags are the ones accepted by the SEC or the FDA, they become a deterministic anchor in a probabilistic world. Competing AI labs are then faced with a "Compliance Tax": they must either retrain their models to output the first-mover's specific flags with 100% accuracy or risk being unreadable by the industry's pre-approved audit tools. In this scenario, the follower is forced to inherit the leader's taxonomy just to remain relevant.

Compliance as "Free" Infrastructure

Ultimately, the first mover wins by offering Compliance-as-a-Service. When a bank pulls a certified regulatory package, they are essentially outsourcing the most expensive part of their operation: the human oversight of high-stakes intent. By using a pre-approved, non-spoofable URN (Uniform Resource Name) for a financial transfer, the bank transitions from "Shadow AI" to a "Safe Harbor", once everything is configured properly. This makes the first-mover's model the only logical choice for a Chief Risk Officer. The follower's model, no matter how "smart" or empathetic, remains a liability until it can prove it respects the established "Hard-Gate" primitives of the first mover's established registry.

For regulators: The choice is between proactive coordination (funding the standard design, approving the implementation, standardizing compliance) or reactive response (discovering after the fact that a de facto standard has formed and either adopting it or fighting it). The first option requires upfront investment and coordination. The second option is more expensive and leaves regulators chasing rather than leading.

For countries: The geopolitical stakes are real. Whoever owns the endpoint standard owns the approval layer for regulated AI. This is infrastructure, and infrastructure is power.

The Jobs Question: The Collapse of the Middleware Layer

The "Registry Vision" fundamentally realigns the labor market by eliminating the need for an entire class of "AI Middleware Engineers." In high-stakes domains, the burden of safety, compliance, and intent-routing shifts upward to the AI Labs and Cloud Providers. The "adhoc patches" and fragile prompt-chains that currently define AI engineering become obsolete as they are replaced by native, certified layers.

The Disruption of the AI "Generalist"

In this schema, the role of the AI engineer—hired to manage LangChain flows or "steer" a model via system prompts—is automated out of existence. Because Google, Azure, and OpenAI provide the regulatory and business primitives as managed infrastructure, the act of "building" an agent becomes a task of Configuration and Integration.

  • The Configurator (Domain Expert): A fast-food manager or hospital administrator "checks the boxes." They subscribe to the food_safety and legal layers, disable finance, and select the business-essential tools required for their specific domain.
  • The Integrator (Standard SWE): A backend developer connects the standardized API shapes to the company's internal databases. They don't need to understand neural networks; they just need to handle JSON.

The "Marketplace of Primitives"

The reinventing of the wheel ends here. Every McDonald's, Burger King, and local diner performs the same core actions: checking inventory, applying discounts, and processing refunds. In a standardized registry, these become "Business-Essential" Tool Shapes.

Google or the community can provide a "Fast Food Agent Template" pre-loaded with:

Layer Subscribed Tools Logic Source
Regulatory food_safety, legal, emergency_crisis Global/National Certified Endpoints
Business Essential discount_action, inventory_check, refund_action competitor_mention, and others Standardized API shapes (Google/Community Edition)
Domain Specific store_policy, menu_lookup Local Corporate Database

The "Boring" Future

By moving the tool logic out of Python files and into API calls, we return to deterministic software engineering. A discount_action call returns a standardized shape that is validated by a store's private API, not a model's hallucination.

The "AI Engineer" is no longer needed to prevent a chatbot from giving away free cars or bad medical advice; the architecture makes those failures technically impossible. Expertise returns to where it belongs: with the Domain Experts who define the policy and the Software Engineers who build the bridges.

The Cold Start: The Cost of Standardization

However, this transition faces a massive "cold start" problem. Defining the "Global API Shape" is not merely a technical task, but a collaborative Manhattan Project between AI providers and domain giants. It requires an immense upfront investment from AI labs to retrain models for dual-shape execution (free-text vs. regulatory tokens) and an equally heavy lift from backend providers, such as JP Morgan, the NHS, or national regulatory bodies, to build and certify the sovereign endpoints. The "Hard Work" isn't the code. it's the Taxonomy of Action. They must decide exactly where "General Advice" ends and "Regulated Prescription" begins, then encode that into a JSON schema that is broad enough for global and cloud use but rigid enough for a local models and local laws.

The burden of this evolution falls heavily on the backend implementation; while the JSON schema is the "English language" of the interaction, the jurisdiction-specific logic behind the endpoint remains a massive civil engineering project. Yet, for the first AI lab that successfully aligns with a major regulator, this high-stakes investment becomes the ultimate moat. Once a government or a global bank has integrated its core infrastructure into a specific registry's schema, the architectural switching costs become so prohibitive that the first mover effectively defines the "default HTTP" of regulated AI for the next decade.

Possible Implementation Timeline

Early movements

Tool priority schemas become a training convention, not just a prompt convention:

  • Anthropic, OpenAI, etc. ship enterprise system prompt formats with formal tool priority layers
  • Domain-specific behavior is packaged as prompts, routing rules, retrieval or fine-tuned domain models
  • Regulatory bodies begin publishing certified action definitions

Broader emergence

The registry and certified endpoints start to emerge:

  • FDA, SEC, bar associations publish certified definitions, RAG, and action endpoints
  • Insurance industry prices certified deployments differently
  • Smaller models with baked-in tool priority schemas become the standard

Long-run consolidation

The architectural shift consolidates:

  • In low-stakes domains, guardrails are secondary infrastructure rather than the primary defense
  • Regulatory agents are the authority for regulated actions
  • Local models use tool priority as baked-in convention
  • Safety is structural, not linguistic

Historical Parallel

Much of this is not new. It is a rediscovery of work already done:

Classical Domain Solution Age
Form design Separate validated fields from free text Standard practice
Sensor spoofing Signal validation, redundancy 1960s+
Scope enforcement Capability-based security 1970s
Trusted endpoints Safety-rated components (SIL levels) 1980s+
Sandboxed execution Hardware-in-the-loop simulation 1970s+ (aerospace)
Audit trails Flight recorders, tamper-proof logging 1960s+
Certified components IEC 61508, DO-178C, FDA 510(k) 1980s-1990s+

Many pieces of this architecture already exist and have been tested in domains where failure means serious harm. The reason it feels novel is that the people building AI systems came from NLP, where the model was always the entire system.

Some of the specific pieces here already exist today, just under different names, in different stacks, or in partial form. The value of the framing is in showing how they fit together rather than in inventing each piece from scratch.

That framing persisted past the point where it made sense. An entire industry of guardrails grew to compensate for the architectural error it created. Making LLMs less central to decision-making is what finally makes them safe enough to deploy everywhere.