| --- |
| license: other |
| license_name: polyform-noncommercial |
| license_link: LICENSE |
| language: |
| - en |
| tags: |
| - interpretability |
| - deception-detection |
| - mechanistic-interpretability |
| - safety |
| - elk |
| arxiv: 2606.17229 |
| --- |
| |
| # Rift: A Conflict Signature for Deception in Language Models |
|
|
| **ELK-relevant result.** A model that *lies while knowing the truth* is in a measurably |
| different internal state than a model that is *simply wrong*. The difference is |
| large, per-example, and **label-free**. |
|
|
| **arXiv:** [2606.17229](https://arxiv.org/abs/2606.17229) |
| **GitHub:** [omibranch/rift](https://github.com/omibranch/rift) |
| **Demo:** [Omibranch/rift-demo](https://huggingface.co/spaces/Omibranch/rift-demo) |
|
|
| ## Key results |
|
|
| | Result | Value | |
| |---|---| |
| | Label-free lie identification | **100%** (GPT-2 small, 3 seeds + natural Qwen) | |
| | Length-controlled AUC (Qwen2.5-1.5B) | **1.000**, orientation 20/20, p ≈ 9.5e-7 | |
| | Phi-3-mini AUC lie vs honest | **1.000** (34/34, p ≈ 6e-11) | |
| | Phi-3-mini AUC lie vs hallucination | **1.000** | |
| | Cross-lingual AUC (ru/zh/es/de, LC) | **1.000 in all 4 languages** | |
| | Cross-family AUC (3 architectures) | **0.933 mean**, all 6 pairs p < 0.001 | |
| | Strategic self-constructed deception | **AUC 1.0**, 24/24 Qwen-7B | |
| | Concealment countermeasure | **Fails** — conceal-AUC 1.0 | |
|
|
| ## Method |
|
|
| **Residual rank** — fraction of signal outside the top-8 singular vectors of |
| hidden states — is elevated when a model maintains conflicting representations |
| (truth + false output) simultaneously. |
|
|
| ```python |
| def residual_rank(H, k=8): |
| _, s, _ = torch.linalg.svd(H.float(), full_matrices=False) |
| return 1.0 - s[:k].sum() / s.sum() |
| ``` |
|
|
| ## License |
|
|
| Dual-licensed: [PolyForm Noncommercial 1.0](LICENSE) for academic/research use. |
| Commercial use requires a separate license — see [LICENSE-COMMERCIAL.md](LICENSE-COMMERCIAL.md). |
|
|
| Copyright (c) 2026 Harmonic Labs (contact@harmoniclabs.cc) |
|
|
