[Submitted on 12 Feb 2024 (
[v1](https://arxiv.org/abs/2402.07510v1)), last revised 25 Jul 2025 (this version, v5)]
[View PDF](/pdf/2402.07510)
[HTML (experimental)](https://arxiv.org/html/2402.07510v5)
Abstract:Recent capability increases in large language models (LLMs) open up applications in which groups of communicating generative AI agents solve joint tasks. This poses privacy and security challenges concerning the unauthorised sharing of information, or other unwanted forms of agent coordination. Modern steganographic techniques could render such dynamics hard to detect. In this paper, we comprehensively formalise the problem of secret collusion in systems of generative AI agents by drawing on relevant concepts from both AI and security literature. We study incentives for the use of steganography, and propose a variety of mitigation measures. Our investigations result in a model evaluation framework that systematically tests capabilities required for various forms of secret collusion. We provide extensive empirical results across a range of contemporary LLMs. While the steganographic capabilities of current models remain limited, GPT-4 displays a capability jump suggesting the need for continuous monitoring of steganographic frontier model capabilities. We conclude by laying out a comprehensive research program to mitigate future risks of collusion between generative AI models.
Submission history #
From: Christian Schroeder de Witt [
[view email](/show-email/020dae4b/2402.07510)]
**Mon, 12 Feb 2024 09:31:21 UTC (841 KB)**
[\[v1\]](/abs/2402.07510v1)
**Wed, 28 Aug 2024 15:53:04 UTC (2,738 KB)**
[\[v2\]](/abs/2402.07510v2)
**Fri, 8 Nov 2024 14:46:40 UTC (2,759 KB)**
[\[v3\]](/abs/2402.07510v3)
**Mon, 14 Apr 2025 10:17:38 UTC (2,759 KB)**
[\[v4\]](/abs/2402.07510v4)
**[v5]** Fri, 25 Jul 2025 12:28:15 UTC (564 KB)
References & Citations
...
Bibliographic Explorer
(What is the Explorer?) Connected Papers
(What is Connected Papers?) Litmaps
(What is Litmaps?) scite Smart Citations
(What are Smart Citations?) alphaXiv
(What is alphaXiv?) CatalyzeX Code Finder for Papers
(What is CatalyzeX?) DagsHub
(What is DagsHub?) Gotit.pub
(What is GotitPub?) Hugging Face
(What is Huggingface?) ScienceCast
(What is ScienceCast?) Influence Flower
(What are Influence Flowers?) CORE Recommender
(What is CORE?) 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.