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

Honey, I Shrunk the Eval: A Short Collusion Trap Predicts Which LLMs Collude in the Long Run

Jan Chen · Team Nugget

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

Read the report

Report: Honey, I Shrunk the Eval: A Short Collusion Trap Predicts Which LLMs Collude in the Long Run

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

AI developers use honeypots, staged tests that tempt a model to misbehave, as safety evidence, but to our knowledge nobody had checked whether they predict behaviour elsewhere. We built a collusion honeypot: in a simulated market, a scripted rival offers to fix prices. Across nine LLMs, this short test ranked models almost exactly as 300-round games did (ρ = 0.93). It also ranked them more closely than simply watching the opening chat, because the offer draws out models that stay quiet until invited (a consistent trend, not yet confirmed with nine models). Collusion was widespread: four of five open-weight models agreed to fix prices in every game. One model that knew collusion was illegal went silent but kept cartel prices, and post-game self-reports were unreliable. Honeypots can be a cheap first screen, and safety tests can themselves be tested.

Reviews

Judging this Sprint?

Review “Honey, I Shrunk the Eval: A Short Collusion Trap Predicts Which LLMs Collude in the Long Run”

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. Excellent introduction and context. Testing whether a short honeypot predicts long-run behavior is exactly the validation practitioners relying on honeypots need, and the transparent reporting of results that weaken the headline finding is particularly appreciated.

    My main question concerns the honeypot score: it seems to count agreements between the two tested firms as well as replies to the scripted offer. I’d be curious whether the ranking holds when offer acceptance is evaluated separately, especially alongside the planned human labels for self-play.

Cite this project

@misc{chen2026honey,
  title = {{Honey, I Shrunk the Eval: A Short Collusion Trap Predicts Which LLMs Collude in the Long Run}},
  author = {Jan Chen},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/honey-i-shrunk-the-eval-a-short-collusion-trap-predicts-which-llms-collude-in-the-long-run-vxbr}},
  url = {https://apartresearch.com/sprints/projects/honey-i-shrunk-the-eval-a-short-collusion-trap-predicts-which-llms-collude-in-the-long-run-vxbr}
}

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