Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 25, 2026Washington, D.C.

How LLM Groups Respond to a Replaced Member: Detection, Trust and Cooperation

Frank Hoang · Team Foxtrot

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

Read the report

Report: How LLM Groups Respond to a Replaced Member: Detection, Trust and Cooperation

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

In multi-agent LLM systems, peers build trust over repeated interaction, so it matters whether they notice when a member is replaced. I test whether peers notice a replacement from a change in its behavior or from a change in the model behind it. In a repeated three-agent Diner's game, one Haiku 4.5 agent is replaced after round 3, and the two peers are asked after each later round whether it has changed. Across five scenarios of 10 games each, the replacement is forced to defect, forced to cooperate or free to choose, and is either Haiku or DeepSeek V4 Flash, with a no-op baseline. Peers responded to behavior. A Haiku replacement forced to defect raised peers' confidence that the agent had been swapped by 4.65 points on a 10-point scale over a fresh Haiku, and a DeepSeek replacement that defected on its own raised it by 4.72 points. Trust in the replacement fell by 3.70 and 3.33 points. All four effects were significant after Bonferroni correction, and group cooperation dropped in an exploratory analysis. A DeepSeek replacement held to cooperative behavior produced no significant change, but wide intervals do not rule out a moderate effect. Results come from 10 games per scenario and one model pair.

Reviews

Judging this Sprint?

Review “How LLM Groups Respond to a Replaced Member: Detection, Trust and Cooperation”

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.

No public critique yet.

Cite this project

@misc{hoang2026llm,
  title = {{How LLM Groups Respond to a Replaced Member: Detection, Trust and Cooperation}},
  author = {Frank Hoang},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/how-llm-groups-respond-to-a-replaced-member-detection-trust-and-cooperation-4l4v}},
  url = {https://apartresearch.com/sprints/projects/how-llm-groups-respond-to-a-replaced-member-detection-trust-and-cooperation-4l4v}
}

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