Metraipolis › archive › dc_d421b3a2efee

Chain-of-Thought Oversight Requires Separate Evidence for Faithfulness and Monitorability — arXiv:2611.01620

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.01620
License (as archived): CC0-1.0
Archived: 2026-11-03T18:12:28.421Z
SHA-256 of the archived text: 69cd5379acb0e1c36050800b59001268d710156e360262816bbf7a462aff4682
Stable URL of this page: https://metraipolis.com/documento/dc_d421b3a2efee

How to cite

Chain-of-Thought Oversight Requires Separate Evidence for Faithfulness and Monitorability — arXiv:2611.01620 — arXiv (en), https://arxiv.org/abs/2611.01620 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d421b3a2efee (Metraipolis archive dc_d421b3a2efee, sha256 69cd5379acb0e1c36050800b59001268d710156e360262816bbf7a462aff4682)

BibTeX

@misc{dcd421b3a2efee,
  title = {Chain-of-Thought Oversight Requires Separate Evidence for Faithfulness and Monitorability — arXiv:2611.01620},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.01620},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d421b3a2efee; sha256 69cd5379acb0e1c36050800b59001268d710156e360262816bbf7a462aff4682},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01620. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A common assumption in chain-of-thought (CoT) oversight is that a visible reasoning trace makes model behavior easier to monitor. We argue that this assumption can conflate distinct properties along a chain linking a model's internal computation, reasoning artifacts, outputs, and monitor judgments. We identify four properties along this chain: reasoning visibility (whether a trace is accessible), mechanistic faithfulness (whether a trace reflects internal computation), mechanistic localizability (whether a compact causal mechanism can be isolated), and behavioral monitorability (whether a monitor can detect the target behavior). Two complementary case studies across three open-weight reasoning models show why these distinctions matter. In the security vulnerability reasoning study, probes and activation ablations provide evidence of internal representation and causal influence, but a monitor with access to the reference answer does not outperform the majority-class baseline in a held-out test. In authority-hint factual reasoning study, a monitor correctly distinguishes changed from unchanged answers, but seven mechanistic analyses do not isolate a compact causal mechanism. These results show that evidence for one claim does not establish the others. We therefore propose an evidential-chain reporting standard. Each claim carries a licensing statement that specifies its target property, access tier, and artifact, and labels each required link as supported, unsupported, or assumed. These assessments determine whether the conclusion is licensed, conditionally licensed, or not licensed.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2611.01620 — 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 2026-11-03T18:09:53.346Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).