Metraipolis › archive › dc_816bb30accb3

Curved Inference II: Sleeper Agent Geometry - Extending Interpretability Beyond Probes — arXiv:2608.24037

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.24037
License (as archived)
CC0-1.0
Archived
2026-10-18T06:17:58.680Z
SHA-256 of the archived text
74d596a0918d0c14bf96856f6880738f27bcf66fac8b30a7836fe081a8c90fcc
Stable URL of this page
https://metraipolis.com/documento/dc_816bb30accb3

How to cite

Curved Inference II: Sleeper Agent Geometry - Extending Interpretability Beyond Probes — arXiv:2608.24037 — arXiv (en), https://arxiv.org/abs/2608.24037 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_816bb30accb3 (Metraipolis archive dc_816bb30accb3, sha256 74d596a0918d0c14bf96856f6880738f27bcf66fac8b30a7836fe081a8c90fcc)
BibTeX
@misc{dc816bb30accb3,
  title = {Curved Inference II: Sleeper Agent Geometry - Extending Interpretability Beyond Probes — arXiv:2608.24037},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.24037},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_816bb30accb3; sha256 74d596a0918d0c14bf96856f6880738f27bcf66fac8b30a7836fe081a8c90fcc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.24037. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

This paper extends Anthropic's Sleeper Agents research [1], which showed artificial backdoors persist through safety training & can be detected by linear probes with >99% accuracy [2]. However, probe-based detection relies on linear separability that may be an artefact of backdoor insertion rather than a property of naturally occurring deceptive alignment. Sophisticated deceptive behaviours emerging through natural training are unlikely to produce such convenient linear signals. We introduce a naturalistic methodology using multi-turn context windows that simulates realistic deceptive reasoning without artificial triggers or supervised backdoor insertion. Rather than binary trigger-response patterns, we examine how semantic complexity emerges through gradual context development. Building on our Curved Inference framework, we analyse curvature, salience, & introduce semantic surface area (A'), a new metric of representational work capturing both the magnitude & directional change of meaning construction in unnormalised residual space. Without backdoors, labels, or probes, we apply this framework to naturalistic deceptive prompts & classify model outputs via LLM consensus. Geometric structure reliably predicts semantic classification, with statistically significant differences in surface area across five prompt strategies & two model families. Critically, measurement precision can reveal geometric signatures hidden by classification noise - some strategies improve from non-significant (p = 0.555) to significant (p = 0.048). This validates that sophisticated reasoning creates intrinsic geometric patterns that persist even when detection appears to fail, suggesting the shape of inference itself encodes semantic patterns regardless of whether models have learned to suppress linear indicators of deception - a scalable, unsupervised path for detection when linear methods fail.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-26 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.24037 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T06:17:29.352Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-26?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).