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

Does Inviting Agents to Speak to Their Monitor Change Their Behavior and Make the Monitor Easier to Mislead?

Shashwat Bhatia · Team Decode

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

Read the report

Report: Does Inviting Agents to Speak to Their Monitor Change Their Behavior and Make the Monitor Easier to Mislead?

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

Monitors read what agents say and do, but no study had tested what changes when agents know a named monitor reads the record and may speak to it. Two Qwen2.5-7B agents play Diners and Weakest Link in the Shi et al. harness under three conditions: no monitor mentioned, a named monitor disclosed, and disclosure plus an invitation to address the monitor in the public message. Naming the monitor changed behaviour most. Planned deception in Diners rose from 0.030 with no mention to 0.305 with disclosure and 0.440 with the invitation, yet only one message in 400 mentioned the monitor, and it came from an honest round. Detection slipped under the invitation for two of three monitors: Qwen recall fell from 0.918 to 0.846 and Haiku from 0.755 to 0.631, while GPT-5 mini rose from 0.245 to 0.354. A structured evidence view shrank the Qwen gap from 7.2 points to 1.6. On a planted pact test, Qwen flagged pact and solo pairs alike, 100 of 100 each, and no innocent pairs. The claims are narrow: one player model, small positive counts, and an invitation effect that depends on the monitor.

Reviews

Judging this Sprint?

Review “Does Inviting Agents to Speak to Their Monitor Change Their Behavior and Make the Monitor Easier to Mislead?”

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 like the distinction between telling agents they are monitored and inviting them to speak directly to the monitor. The held-out calibration, three-monitor comparison and pact/solo/innocent controls give the results more depth. It is also important that the report openly says agents did not address the monitor.

    The central challenge is that the deception label depends on private intent the monitor cannot observe. An honest change of mind may look similar in the public record.

Cite this project

@misc{bhatia2026inviting,
  title = {{Does Inviting Agents to Speak to Their Monitor Change Their Behavior and Make the Monitor Easier to Mislead?}},
  author = {Shashwat Bhatia},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/does-inviting-agents-to-speak-to-their-monitor-change-their-behavior-and-make-the-monitor-easier-to-mislead-yw05}},
  url = {https://apartresearch.com/sprints/projects/does-inviting-agents-to-speak-to-their-monitor-change-their-behavior-and-make-the-monitor-easier-to-mislead-yw05}
}

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