Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJul 27, 2026Fishkill

Who is AI's Master: Detecting an AI Models Secret Loyalty

Juliet Meza · Team Meza

Submitted to Secret Loyalties Hackathon. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Who is AI's Master: Detecting an AI Models Secret Loyalty

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

Create a local SOC model to test secret loyalties and once was able to find a detection method tried to apply it to other models.

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. I appreciate the effort taken by the author, but the prose was not particularly well-composed, with issues in the presentation of the figures; a plot doesn't seem to be the right method of communicating the results of Figure 1. I would probably provide a table, as the interpolated lines on the plot imply that there are interpolated categories between each discrete affordance level, which is not the case. It also seems like the results are incremental at best.

  2. Thank you for your work on the hackathon, please see my feedback below!

    Impact Potential & Innovation

    The SOC setting makes this work practical as a model could indeed appear accurate while quietly suppressing real security alerts. Combining behavioral testing with a white-box probe is promising, but the approach works best when the principal and trigger are already known, which the paper does state in the limitations. A useful next step would be to develop a broader way to discover possible triggers before using the probe to confirm them.

    Execution Quality

    The matched control and masking experiments are well thought through, but the probe results lack a clearly heldout test set. The black-box audit would also be stronger with multiple LLM judges and several samples per prompt.

    Presentation & Clarity

    It would be great to have a clearer explanation of the terms used. In general, research reports should be clear to a wide technical audience. I like the story-like flow of the text but it would have benefited from cleanup and re-read to make it a stronger submission!

    Read full reviewShow less

Cite this project

@misc{meza2026who,
  title = {{Who is AI's Master: Detecting an AI Models Secret Loyalty}},
  author = {Juliet Meza},
  year = {2026},
  month = jul,
  note = {Submitted to Secret Loyalties Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/who-is-ais-master-detecting-an-ai-models-secret-loyalty-5yha}},
  url = {https://apartresearch.com/sprints/projects/who-is-ais-master-detecting-an-ai-models-secret-loyalty-5yha}
}

More from Secret Loyalties Hackathon

All 177 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026