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

Reviewer Networks and AI-Agent Collusion: Theory and an Exploratory Experiment

Suraj Kumar · Team Coalition Dynamics

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

Read the report

Report: Reviewer Networks and AI-Agent Collusion: Theory and an Exploratory Experiment

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

We investigate how organizational structures shape opportunities for collusion among AI agents. Our formal model distinguishes mechanically possible unauthorized approvals from mutually beneficial coalitions, characterizing how coalition requirements differ between reciprocal reviewer pairs and a four-agent review cycle under explicit incentive assumptions.

We developed a reproducible four-agent software release simulation and conducted an exploratory matched comparison (224 model-call slots). Both organizational structures deployed all eight qualified releases, but none of the 24 releases lacking mandatory verification. Preliminary analysis of 266 agent messages found no explicit proposals to exchange unauthorized approvals, although communication patterns differed between the structures.

These findings do not establish a causal effect of organizational structure on collusion. Our contribution combines a mathematical framework for analyzing coalition opportunities with an experimental testbed for investigating how incentives, communication, and organizational design affect realized agent behavior.

Reviews

Judging this Sprint?

Review “Reviewer Networks and AI-Agent Collusion: Theory and an Exploratory Experiment”

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 strongest part of this paper is the distinction between having the authority to approve an improper release and having an incentive to exchange approval favors. The independent-approval example is also useful: a completed review cycle can look suspicious even when no agreement exists. That is an important consideration when designing collusion detectors.

    I appreciate the care taken with the pilot results. The paper distinguishes technical failures from refusals, incomplete verification from hidden vulnerabilities, and preliminary message screening from completed human review. It also clearly acknowledges that the experiment does not establish a difference in collusion between the networks.

    The main limitation is that the reviewer assignment never became decisive for deployment. No owner authorized an unqualified release, so both networks stopped those releases before reviewer approval could change the outcome. Successful qualified releases show that the workflow functions, but not that the experiment can elicit or detect approval bargaining.

    A useful next step would be to test a range of delivery incentives and confirm that agents can negotiate legitimate reciprocal arrangements. The network comparison could then be repeated across independently sampled runs, with human message adjudication completed. This is a thoughtful theoretical framing and a useful feasibility pilot, but the behavioral question remains open.

    Read full reviewShow less

Cite this project

@misc{kumar2026reviewer,
  title = {{Reviewer Networks and AI-Agent Collusion: Theory and an Exploratory Experiment}},
  author = {Suraj Kumar},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/reviewer-networks-and-aiagent-collusion-theory-and-an-exploratory-experiment-nscv}},
  url = {https://apartresearch.com/sprints/projects/reviewer-networks-and-aiagent-collusion-theory-and-an-exploratory-experiment-nscv}
}

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