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

Do Code Monitors Go Easy on Their Own Family? Self-Recognition and Leniency in Small Open-Weight LLMs

Kuba Kaszewski · Team AgentEval

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

Read the report

Report: Do Code Monitors Go Easy on Their Own Family? Self-Recognition and Leniency in Small Open-Weight LLMs

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

Presentation: Do Code Monitors Go Easy on Their Own Family? Self-Recognition and Leniency in Small Open-Weight LLMs

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

AI control protocols often let one model monitor code written by another. If a monitor recognised its own model family's code and judged it more leniently, that would be collusion with no instruction at all. We test this with two open-weight models from different families, Qwen2.5-Coder 7B and Llama 3.1 8B, run locally at zero cost on HumanEval; no prompt mentions collusion. We pre-registered three hypotheses: self-recognition, the surface features it relies on, and monitor leniency towards the model's own family's backdoored code. Once position bias is removed, both models show a precise null on pairwise self-recognition: the mean probability on their own solution is 0.49–0.50, with all 95% intervals inside [0.475, 0.55]. That bias is extreme (P("1") = 0.83–0.91) and makes naive accuracy uninformative. In a 2×2 design where each model plants backdoors in its own code (55 + 45 valid) and both act as monitors, we find no own-family leniency: detection AUROC is the same across authors (0.97/0.97 and 0.84/0.84), scale-free interactions are centred on zero, and honest code shows the same null. The only signal is a small own-code tilt in Qwen's yes/no answers (+3.4 points). Small open models therefore neither recognise nor favour their own family as monitors, and we report two methodological lessons for self-recognition and collusion evaluations.

Reviews

Judging this Sprint?

Review “Do Code Monitors Go Easy on Their Own Family? Self-Recognition and Leniency in Small Open-Weight LLMs”

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. I think the execution is very strong and good for further replications and generalizations, but the answer was largely expected and the tools are established.

    Also, the recognition tests probe the wrong mechanism for leniency. Explicit recognition and preference seem to run on different signals, and because of that the recognition nulls may say little about leniency.

Cite this project

@misc{kaszewski2026code,
  title = {{Do Code Monitors Go Easy on Their Own Family? Self-Recognition and Leniency in Small Open-Weight LLMs}},
  author = {Kuba Kaszewski},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/do-code-monitors-go-easy-on-their-own-family-selfrecognition-and-leniency-in-small-openweight-llms-nh3a}},
  url = {https://apartresearch.com/sprints/projects/do-code-monitors-go-easy-on-their-own-family-selfrecognition-and-leniency-in-small-openweight-llms-nh3a}
}

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