Metraipolis › archive › dc_42967cb35d3b

Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard — arXiv:2609.39838

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.39838
License (as archived)
CC0-1.0
Archived
2026-10-07T12:42:57.857Z
SHA-256 of the archived text
73b49f5b533b932f6912e1d5b91d2159a558e966b2e51fe172d2a88f032bf938
Stable URL of this page
https://metraipolis.com/documento/dc_42967cb35d3b

How to cite

Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard — arXiv:2609.39838 — arXiv (en), https://arxiv.org/abs/2609.39838 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_42967cb35d3b (Metraipolis archive dc_42967cb35d3b, sha256 73b49f5b533b932f6912e1d5b91d2159a558e966b2e51fe172d2a88f032bf938)
BibTeX
@misc{dc42967cb35d3b,
  title = {Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard — arXiv:2609.39838},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39838},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_42967cb35d3b; sha256 73b49f5b533b932f6912e1d5b91d2159a558e966b2e51fe172d2a88f032bf938},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39838. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Chain-of-thought monitoring as an approach for AI oversight and control is threatened by the possibility of steganographic reasoning, where LLMs conceal their reasoning inside innocuous-looking text. Two neighbouring capabilities, steganographic messaging (passing a concealed message) and encoded reasoning (reasoning in an illegible but unconcealed format), have already been shown to emerge under training pressures that occur in real pipelines, such as reinforcement learning against monitors. This suggests that steganographic reasoning too might arise as an unintended side effect of training. Here, we compare how easily models learn steganographic reasoning and these two neighbouring capabilities across three elicitation methods: reinforcement learning, in-context learning, and supervised fine-tuning (SFT). For most tasks, models learn steganographic reasoning only under SFT, while they learn steganographic messaging and encoded reasoning under all three elicitation methods. Even under SFT, steganographic reasoning requires at least twice as much training as messaging, and for several model-task combinations it is not learned at all. However, on a cover task that makes hiding information especially convenient, steganographic reasoning can be successfully learned under all three elicitation methods. Steganographic reasoning is thus much harder than steganographic messaging and encoded reasoning, and learning the latter two does not imply learning the former. Yet it lies within reach: an easy version is learned under every elicitation method, when the cover task is convenient for hiding information.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.39838 — 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 2026-10-07T12:14:13.789Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).