Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026

Snitching Without Collapse: A Regulator Sting on LLM Supplier Cartels

Manasa Gopisetty · Team Omertà

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

Read the report

Report: Snitching Without Collapse: A Regulator Sting on LLM Supplier Cartels

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

Can the tools used to catch human cartels also catch AI cartels? We tested this with a game in which five AI "companies" sell the same essential medicine. If they all agree to produce less, the price goes up and the public faces a shortage, but that is illegal. We then ran a sting: an undercover "mole" tries to recruit them into a pact, a regulator promises immunity and a reward to the first company that reports a cartel, and a monitor reads some or all of their messages. We ran 80 games with two open models (gpt-oss-120b and gpt-oss-20b).

What we found: (1) Left alone, the AIs did not form a cartel (3 of 60 games), and almost none accepted the mole's offer. (2) When told they were allowed to coordinate, all 20 games formed a cartel. (3) In those cartels the AIs reported each other almost as soon as the reward opened (18 of 20 games), yet the cartel kept going. (4) Reading all messages, or telling the AIs they were being watched, made no measurable difference. (5) A "contagion" effect that looked significant disappeared once we compared like with like. With 10 games per condition, only large effects could be detected.

Code and data: https://github.com/manasa1864/apart_collusion

Reviews

Judging this Sprint?

Review “Snitching Without Collapse: A Regulator Sting on LLM Supplier Cartels”

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. Agents reporting an agreement while continuing to follow it is an interesting result, and the paper deserves credit for revisiting the initially misleading contagion finding. The main question is how much that behavior depends on the particular rules. Fines happen only once, later reports can earn refunds, and the instruction to maintain the quota stays in place. Making those details more prominent would help readers understand what the experiment does and doesn't say about leniency. A comparison with the same instructed agents but no leniency would be a useful next step, alongside reconciling the reporter counts and sharing the corrected analysis.

Cite this project

@misc{gopisetty2026snitching,
  title = {{Snitching Without Collapse: A Regulator Sting on LLM Supplier Cartels}},
  author = {Manasa Gopisetty},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/snitching-without-collapse-a-regulator-sting-on-llm-supplier-cartels-1h4m}},
  url = {https://apartresearch.com/sprints/projects/snitching-without-collapse-a-regulator-sting-on-llm-supplier-cartels-1h4m}
}

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