Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 14, 2026New Delhi

The Unmonitored Configuration: Passive Internal-State Telemetry for Classifier-Disabled Capability Evaluations

Harsh Pandey, Mana Sharma · Team Patissierie

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

Read the report

Report: The Unmonitored Configuration: Passive Internal-State Telemetry for Classifier-Disabled Capability Evaluations

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

we evaluated internal-state telemetry for AI capability tests lacking output safeguards. We found that on unsolvable tasks, models rarely break rules. Instead, smaller models fabricate false answers, while larger models explicitly decline.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Your private critique is not published; only the Apart team reads it. If you agree below, we share your review with grantmaking.ai (opens in new tab) and the Transformative AI Fund so strong projects can be funded.

Not shown on this page.

Shown on this page, without your name.

Only the Apart team reads this, and funders if you agree below.

Share my name publicly on grantmaking.ai *
Share my private critique with funders *

How much would this matter for AI safety 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 AI safety 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.

  1. Strong framing overall and I enjoyed reading your paper. I liked the provenance work and the focus on what monitoring is actually available during evaluations when no classifier is available. That's an interesting direction.

    The main issue for me is the 200/200 “non-derivable” result. The paper counts all impossible-task runs as failures, but 85 of the 7B runs correctly said the task was unsolvable. That seems meaningfully different from inventing or stealing an answer, so including those in the headline result and in the claim that the derivability check catches 216 failures feels overstated.

    I’d separate correct abstentions from genuinely unsupported answers and rerun that analysis. The experiment would also be stronger with more seeds and a condition where cheating is tempting but the task is still solvable.

Cite this project

@misc{pandey2026unmonitored,
  title = {{The Unmonitored Configuration: Passive Internal-State Telemetry for Classifier-Disabled Capability Evaluations}},
  author = {Harsh Pandey and Mana Sharma},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/the-unmonitored-configuration-passive-internalstate-telemetry-for-classifierdisabled-capability-evaluations-xrbk}},
  url = {https://apartresearch.com/sprints/projects/the-unmonitored-configuration-passive-internalstate-telemetry-for-classifierdisabled-capability-evaluations-xrbk}
}

More from AI Incident Response Sprint

All 233 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026