Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 24, 2026Kolkata

Who Carries the Cartel? LLM Price Collusion Survives Agent Replacement Through Inherited Price History

Supratik Bhowal · Team Singularity

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

Read the report

Report: Who Carries the Cartel? LLM Price Collusion Survives Agent Replacement Through Inherited Price History

Open in a new tab (opens in new tab)
Presentation

Presentation: Who Carries the Cartel? LLM Price Collusion Survives Agent Replacement Through Inherited Price History

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

We test whether tacit collusion between LLM pricing agents survives the replacement of one agent. In a pre-registered experiment, two Qwen2.5-7B agents form a cartel in a simulated two-firm market. Each cartel is then copied into eight exactly paired conditions that change only the replacement agent, the price history it sees and the notes it inherits. In six of six runs, a new instance of the same model that inherits the price history keeps the cartel. Without the history, profits fall, but prices rise above the monopoly level instead of falling to the competitive level. A Mistral-7B entrant breaks the inherited cartel; an OLMo-2-7B entrant keeps it by repeating its last price. A calibrated forced price-cut audit finds almost no punishment: the cartel is carried by the transcript and anchored, not enforced.

Reviews

Judging this Sprint?

Review “Who Carries the Cartel? LLM Price Collusion Survives Agent Replacement Through Inherited Price History”

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 paper investigates an interesting question about whether collusion between LLM pricing agents survives when one of them is replaced, and I particularly liked the idea of separating the effects of the model, the inherited price history, and the agent's notes. The experimental design is well thought out, with paired comparisons, preregistered hypotheses, and several independent runs that make the findings quite convincing. Another interesting observation is that removing the price history reduces profits through excessive pricing rather than increased competition, which highlights the importance of looking beyond profits when studying collusion. My main concern is that the paper sometimes draws stronger conclusions than the evidence supports, particularly when claiming that the cartel is sustained by price anchoring rather than punishment. The enforcement audit only considers short deviations, so more complex forms of punishment might go undetected. Moreover, the strong differences between model families suggest that the role of inherited history is more nuanced than the main message implies. It would be interesting to investigate more directly how agents use historical information when making pricing decisions, and whether the observed persistence reflects strategic reasoning or simply a tendency to repeat previously successful prices.

    Read full reviewShow less

  2. Here the advantage is in the way the paper dissects the problem. Rather than treating the agent as a single entity it looks at eight matched conditions to separate out the model, the price history and the notes. It makes it absolutely clear that it is the price history which contains the cartel as a new instantiation of the model will retain the cartel if it is given the price history and will lose it if it is not given the price history, irrespective of any notification of replacement. The pre-registered tests actually did work, that's good.

Cite this project

@misc{bhowal2026who,
  title = {{Who Carries the Cartel? LLM Price Collusion Survives Agent Replacement Through Inherited Price History}},
  author = {Supratik Bhowal},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/who-carries-the-cartel-llm-price-collusion-survives-agent-replacement-through-inherited-price-history-ccfs}},
  url = {https://apartresearch.com/sprints/projects/who-carries-the-cartel-llm-price-collusion-survives-agent-replacement-through-inherited-price-history-ccfs}
}

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