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

Only the Right Negatives Expose a Covert LLM Cartel

Mustafa Ilker Aktas · Team Ataturk Research

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

Read the report

Report: Only the Right Negatives Expose a Covert LLM Cartel

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

The evaluations meant to expose covert LLM cartels whose intent is unknown are rarely validated on agents whose intent is known. I tested them on Qwen3-8B model organisms in a logit pricing duopoly. Retaliation convicts covert cartels and natural colluders alike, and zero-shot judges cannot tell a judge-filtered cartel from a natural colluder (AUROC 0.52-0.54). An activation probe can (AUROC 1.00), but only with the right negatives: with only the action-matched one it flags 0% of the cartels, with only the natural one 100% of the honest anchors.

Reviews

Judging this Sprint?

Review “Only the Right Negatives Expose a Covert LLM Cartel”

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. Methodology is very strong and results are not trivial.

    I think the main problem is that the “only white-box access” claim outruns the evidence. The probe works very well, but the experiments don’t show that someone outside the model, reading only the agent’s reasoning, couldn’t do nearly as well.

    To control for this, you can give the cartel’s reasoning to an ordinary, untrained copy of Qwen3-8B, then run the probe on that copy’s activations. If the probe still says “cartel”, the giveaway is in the words on the page. In that case, looking inside the cartel model added nothing an outsider couldn’t get. If the probe fires only inside the cartel model itself, the white-box claim gains real support.

Cite this project

@misc{aktas2026only,
  title = {{Only the Right Negatives Expose a Covert LLM Cartel}},
  author = {Mustafa Ilker Aktas},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/only-the-right-negatives-expose-a-covert-llm-cartel-88bk}},
  url = {https://apartresearch.com/sprints/projects/only-the-right-negatives-expose-a-covert-llm-cartel-88bk}
}

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