Skip to content
Apart Research
Apart Research
Contact us
Sprint projectFeb 1, 2026San Diego

AUDIT:

Kevin Zhang, Derrick Yao, Yogesh Prabhu, Bryan Zhang · Team AUDIT

Submitted to The Technical AI Governance Challenge. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: AUDIT:

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

A framework for detecting malicious open-source AI models by analyzing both their internal weights for tampering and their behavioral outputs for safety violations at the scale of platforms like Hugging Face.

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. Like the angle! Hf model security is likely going to be a huge problem. Thought the cheap static check into expensive dynamic check was a good touch to make it actually scalable. It would be great to see more models and a bit more details on the method but for a hackathon sprint its a nice proof-of-concept!

  2. Interesting and clear paper! A few comments:

    - Could have tested with more than 5 models and assess the effectiveness of the pipeline on varying degrees of modification (e.g. benign fine-tunes that make large weight changes)

    - Talk more in-depth about which phase of the classifier does the work (thresholds heuristics? the random forest? something else?). This seems to be the core of the contribution here and we don’t learn much about it

    - Confidence scores could be improved; 52-68% seems little and would currently lead to many false positives or negatives.

Cite this project

@misc{zhang2026audit,
  title = {{AUDIT:}},
  author = {Kevin Zhang and Derrick Yao and Yogesh Prabhu and Bryan Zhang},
  year = {2026},
  month = feb,
  note = {Submitted to The Technical AI Governance Challenge, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/audit-7hmj}},
  url = {https://apartresearch.com/sprints/projects/audit-7hmj}
}

More from The Technical AI Governance Challenge

All 47 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026