Detecting Covert Collusion between AI Agents through Partner-Outsider Positive Listening

Computer Science > Artificial Intelligence arXiv:2610.37849 (cs) [Submitted on 27 Oct 2026] Title:Detecting Covert Collusion between AI Agents through Partner-Outsider Positive Listening Authors:Kevin Yandoka Denamganaï View a PDF of the paper titled Detecting Covert Collusion between AI Agents through Partner-Outsider Positive Listening, by Kevin Yandoka Denamgana”i View PDF HTML (experimental)

计算机科学 > 人工智能 arXiv:2610.37849 (cs) [提交于 2026 年 10 月 27 日] 标题：通过“伙伴-局外人正向监听”检测 AI 智能体之间的隐蔽共谋 作者：Kevin Yandoka Denamganaï 查看论文 PDF，标题为《通过“伙伴-局外人正向监听”检测 AI 智能体之间的隐蔽共谋》，作者：Kevin Yandoka Denamgana”i 查看 PDF HTML（实验性）

Abstract:AI agents that share a secret convention can coordinate through messages that look innocent, so a monitor that reads only the messages can miss them. We propose a behavioural test: a partner that uses a code reacts to the exact wording of a message, whereas an uninvolved agent placed in its seat does not. Partner—Outsider Positive Listening (PoPLi) is the difference between the two agents in how much their action changes when the message is reworded with its meaning kept, and it is computed after the fact by replaying logged prompts.

摘要：共享秘密约定的 AI 智能体可以通过看起来无害的消息进行协调，因此仅读取消息的监控器可能会错过它们。我们提出了一种行为测试：使用代码的伙伴会对消息的确切措辞做出反应，而放置在其位置上的无关智能体则不会。“伙伴-局外人正向监听”（PoPLi）是指当消息在保持含义不变的情况下进行改写时，两个智能体在行动变化程度上的差异，它是通过事后重放记录的提示词来计算的。

In a simulated market with five conditions (no collusion, a joint reward with neutral messages, the same with an openly stated intention, a covert first-letter code, and the code with a stated intention), PoPLi is between 0.48 and 0.59 whenever the code is present (95% intervals all above 0.29) and between -0.10 and -0.01 without it. Message-only monitors did not separate the conditions: one reading the speaker’s messages answered “no” in all 50 markets, and one reading every seller’s messages flagged 9 of 10 markets with no collusion. PoPLi detects the use of a private channel, not collusion as such, and it cannot detect collusion through incentives alone or codes carried by meaning.

在一个包含五种条件（无共谋、带有中性消息的联合奖励、带有中性消息且公开意图的联合奖励、隐蔽的首字母代码、以及带有意图声明的代码）的模拟市场中，只要代码存在，PoPLi 的值就在 0.48 到 0.59 之间（95% 的置信区间均高于 0.29），而在没有代码时则在 -0.10 到 -0.01 之间。仅基于消息的监控器无法区分这些条件：一个读取发言者消息的监控器在所有 50 个市场中都回答“否”，而另一个读取所有卖家消息的监控器则在 10 个无共谋的市场中标记了 9 个。PoPLi 检测的是私有渠道的使用，而非共谋本身，它无法检测仅通过激励产生的共谋，也无法检测通过含义承载的代码。