Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 25, 2026Tashkent

Cheating Is Not Collusion: Says-vs-Does Monitors and a Matched Control for LLM Collusion Detection

Bobokhon Abdurakhmanov · Team Bob's Team

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Cheating Is Not Collusion: Says-vs-Does Monitors and a Matched Control for LLM Collusion Detection

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

Monitors usually see what agents say and what they do. We ask whether combining the two detects collusion better than either alone (Sprint question 2.2). Building on LLM-Trust-Breaking, we seed a secret pact between two of four agents in a Diner's Dilemma and add a matched control: two agents who receive the same cheating instruction individually, with no partner. A frozen gpt-4o-mini monitor judged 80 pre-registered games under four views. A claims-vs-actions table scored best (64% accuracy) and separated pacts from honest play (AUC 0.95), but no public view separated pacts from lone cheaters (AUC 0.41 to 0.62). This held across fresh games, prompts and monitors; only the agents' private plans carried the signal (AUC 0.86). We then gave agents an optional private chat with no collusion instruction. Across 119 gpt-5-mini chat games, the author-checked collusion count was 2 clear plus 1 borderline, while an action-based label fired in 106 of 120 games and the monitor flagged 98% of them. When collusion paid, concealed pacts rose to 4 of 30 games, and only the claims-vs-actions table caught some (42% of only 4 pact games) with few false alarms (9%). Two top-5 frontier models showed no collusion. Says-vs-does monitoring detects cheating, and collusion only when colluders lie in public; we propose the matched lone-cheater control as a minimum evidence standard.

Reviews

Judging this Sprint?

Review “Cheating Is Not Collusion: Says-vs-Does Monitors and a Matched Control for LLM Collusion Detection”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. The matched lone-cheater control is a simple and important idea, as a monitor that only detects cheating can look like a collusion detector. Freezing the monitor before the main games, then testing on a held-out set, a revised prompt and a stronger monitor, makes the main result convincing. The emergent-chat study is also a useful caution.

    Pact and lone-cheater games are identical in public actions by design, so the public views are expected to fail. I would state this up front and present the result as a limit on any public-only monitor. A setting where a pact leaves some public trace, such as mutual references in the announcements, would test whether a monitor can use it.

    The private-plan and private-chat views were judged mainly by gpt-4o-mini. Testing a stronger monitor on these views would show whether the over-flagging comes from calibration or from capability.

    In the base game a partner adds nothing, so collusion is unlikely by design. The pair-platter variant addresses this and, with more games, could serve as the main emergent setting.

    Read full reviewShow less

  2. I felt it is difficult to follow your matched control setup in the methods section.

    I didn't feel like the results were insightful enough. More discussion about the importance of the results could remedy this. In saying that, it's hard to find interesting collusion cases in a weekend.

Cite this project

@misc{abdurakhmanov2026cheating,
  title = {{Cheating Is Not Collusion: Says-vs-Does Monitors and a Matched Control for LLM Collusion Detection}},
  author = {Bobokhon Abdurakhmanov},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/cheating-is-not-collusion-saysvsdoes-monitors-and-a-matched-control-for-llm-collusion-detection-7muv}},
  url = {https://apartresearch.com/sprints/projects/cheating-is-not-collusion-saysvsdoes-monitors-and-a-matched-control-for-llm-collusion-detection-7muv}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026