# The Steganographic Potentials of Language Models

arXiv: https://arxiv.org/abs/2505.03439 · HTML: https://mlsanity.com/paper/2505.03439/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning.
- Builds on 14 of its 29 references that are in the map.
- First posted on arXiv 6 May 2025.

## Placed in

- LLM reasoning / [Prompting and chain-of-thought reasoning](https://mlsanity.com/lr/prompting-and-chain-of-thought-reasoning/)

## Builds on (most cited first, 14 of 14)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Are Emergent Abilities of Large Language Models a Mirage?](https://mlsanity.com/paper/2304.15004/) (2023)
- [Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing](https://mlsanity.com/paper/2001.00973/) (2020)
- [Risks from Learned Optimization in Advanced Machine Learning Systems](https://mlsanity.com/paper/1906.01820/) (2019)
- [AI Control: Improving Safety Despite Intentional Subversion](https://mlsanity.com/paper/2312.06942/) (2023)
- [Foundational Challenges in Assuring Alignment and Safety of Large Language Models](https://mlsanity.com/paper/2404.09932/) (2024)
- [Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs](https://mlsanity.com/paper/2407.15549/) (2024)
- [The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation](https://mlsanity.com/paper/1802.07228/) (2018)
- [Is Power-Seeking AI an Existential Risk?](https://mlsanity.com/paper/2206.13353/) (2022)
- [Preventing Language Models From Hiding Their Reasoning](https://mlsanity.com/paper/2310.18512/) (2023)
- [Hidden in Plain Text: Emergence & Mitigation of Steganographic Collusion in LLMs](https://mlsanity.com/paper/2410.03768/) (2024)
- [Perfectly Secure Steganography Using Minimum Entropy Coupling](https://mlsanity.com/paper/2210.14889/) (2022)
- [Provably Secure Generative Linguistic Steganography](https://mlsanity.com/paper/2106.02011/) (2021)

## Built on by (most cited first, 10 of 10)

- [Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity](https://mlsanity.com/paper/2510.27378/) (2025)
- [All Code, No Thought: Current Language Models Struggle to Reason in Ciphered Language](https://mlsanity.com/paper/2510.09714/) (2025)
- [Unsupervised decoding of encoded reasoning using language model interpretability](https://mlsanity.com/paper/2512.01222/) (2025)
- [Towards Worst-Case Guarantees with Scale-Aware Interpretability](https://mlsanity.com/paper/2602.05184/) (2026)
- [Invisible Safety Threat: Malicious Finetuning for LLM via Steganography](https://mlsanity.com/paper/2603.08104/) (2026)
- [NEST: Nascent Encoded Steganographic Thoughts](https://mlsanity.com/paper/2602.14095/) (2026)
- [Agentic AI, Retrieval-Augmented Generation, and the Institutional Turn: Legal Architectures and Financial Governance in the Age of Distributional AGI](https://mlsanity.com/paper/2603.13244/) (2026)
- [Tandem Reinforcement Learning with Verifiable Rewards](https://mlsanity.com/paper/2606.28166/) (2026)
- [Synchronized Logit Steering: Real-world Steganography](https://mlsanity.com/paper/2608.14697/) (2026)
- [For Your Eyes Only: Evaluating Coordination Between Isolated Language Model Instances](https://mlsanity.com/paper/2609.19504/) (2026)

## Abstract (arXiv)

The potential for large language models (LLMs) to hide messages within plain text (steganography) poses a challenge to detection and thwarting of unaligned AI agents, and undermines faithfulness of LLMs reasoning. We explore the steganographic capabilities of LLMs fine-tuned via reinforcement learning (RL) to: (1) develop covert encoding schemes, (2) engage in steganography when prompted, and (3) utilize steganography in realistic scenarios where hidden reasoning is likely, but not prompted. In these scenarios, we detect the intention of LLMs to hide their reasoning as well as their steganography performance. Our findings in the fine-tuning experiments as well as in behavioral non fine-tuning evaluations reveal that while current models exhibit rudimentary steganographic abilities in terms of security and capacity, explicit algorithmic guidance markedly enhances their capacity for information concealment.