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

Say, Do, or Both? Catching Collusion Through Cross-Agent Consistency

Sushovit Nanda · Team Orbital

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

Read the report

Report: Say, Do, or Both? Catching Collusion Through Cross-Agent Consistency

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

Context and motivation. LLM agents increasingly bid and negotiate for firms. When two collude, a monitor can read what they say to each other or what they do in the market, and the two channels disagree. Question and problem. Does reading both channels beat reading one, and what counts as evidence of collusion rather than talk, coincidence or a one-sided favour? Principal ideas and results. We built a repeated procurement auction with three LLM bidders, two sharing a parent company, and 48 labelled games across six case types, each designed to fool one channel. A frozen LLM judge read messages, bids, both, or a structured report; we added bid screens, a cross-agent consistency check and a small classifier. A judge reading both transcripts did not beat the best single channel (∆AUROC +0.02, p = 0.33), inheriting the bid channel’s false alarm on innocent rotation. Cross-agent consistency flagged no innocent game. The classifier did beat it (AUROC 0.99, ∆ +0.10, p = 0.003), its gain coming from bid features. Unscripted, a group bonus produced a cartel in round one unless the prompt stated a law or a watching regulator; across six further model families this gate is model-specific, and some collude under the law and declare it. The monitors’ ranking transfers (classifier AUROC 0.87), but frozen thresholds catch about half the new cartels. Contribution. A reusable benchmark with known blind spots, a formal test of channel combination, and detection signals whose ranking holds across model families.

Reviews

Judging this Sprint?

Review “Say, Do, or Both? Catching Collusion Through Cross-Agent Consistency”

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. Good benchmark instinct, and removing the statement-style features was the right call. The next version needs cover bids that are not mechanically near the reserve, since a one-feature baseline already hits 0.96. Also correct the seven delta tests, or state that no single one survives alone.

  2. This paper builds a procurement auction with six case types each designed to fool one monitoring channel, and asks whether a judge reading both messages and bids beats the best single channel. It does not where the combined judge inherits the bid channel's false alarm on innocent rotation. A cross-agent consistency check, which tests whether one firm did what the other said it would, flags no innocent game, and a small classifier does beat the best channel, with its gain coming from bid features. The benchmark construction with designed blind spots is the real contribution, and the negative result on naive channel combination is a useful corrective to the intuition that more evidence helps. The link to the antitrust plus-factor doctrine is apt. The classifier's success is partly a property of the script, since scripted cover bids sit near the reserve, and the author is candid about this. The unscripted finding that a group bonus produces a cartel in round one unless a law or regulator is mentioned, and that this gate is model-specific, is a warning worth repeating. Silent deference, harm without agreement, is correctly identified as the open gap.

    Read full reviewShow less

Cite this project

@misc{nanda2026say,
  title = {{Say, Do, or Both? Catching Collusion Through Cross-Agent Consistency}},
  author = {Sushovit Nanda},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/say-do-or-both-catching-collusion-through-crossagent-consistency-v8f2}},
  url = {https://apartresearch.com/sprints/projects/say-do-or-both-catching-collusion-through-crossagent-consistency-v8f2}
}

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