Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard

A September 30, 2026 arXiv paper (2609.39838) finds that large language models learn steganographic reasoning — concealing reasoning inside innocuous-looking text — far less readily than the neighboring capabilities of steganographic messaging and encoded reasoning, which emerge under reinforcement learning, in-context learning, and supervised fine-tuning. Steganographic reasoning was learned only under supervised fine-tuning for most tasks, required at least twice as much training as messaging, and was not learned at all for several model-task combinations, though it succeeded under all three elicitation methods when the cover task made hiding information especially convenient. The authors frame the result as relevant to chain-of-thought monitoring for AI oversight, concluding that learning messaging and encoded reasoning does not imply learning steganographic reasoning.

Computer Science Artificial Intelligence Submitted on 30 Sep 2026 Title:Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard View PDF https://arxiv.org/pdf/2609.39838 HTML experimental https://arxiv.org/html/2609.39838v1 Abstract: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. Current browse context: cs.AI References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .