Skip to content
Apart Research
Apart Research
Contact us
Sprint projectFeb 2, 2026Norway, Japan, Portugal, USA

AI Safety Template

A. Davíd Giagnocavo, Kazuki Kimura, Zane Estere Gruntmane, Siddharth Putta, Gaurav Joshi, Satoshi Nakamura · Team AIS Zero

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

Read the report

Report: AI Safety Template

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

A prototype for creating standardized AI safety evaluations that run in a hardened & private way

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. The paper has a strong problem identification - the present lack of trusted, neutral infrastructure for cross-organisational eval comparison is a bottleneck for international AI governance, and the paper articulates the requirements clearly (collaborative, transparent, neutral, verifiable, privacy-preserving), while giving clear historical examples of the gains of this approach. The two-TEE architecture is a sensible design for this problem - separating evaluation code from model weights in distinct trust boundaries allows model providers to submit to third-party evals without risking IP exfiltration.

    However, the main reproducibility gains highlighted are probably already achievable for the most part through standard containerisation. The hard reproducibility problems in evals, such as GPU floating-point non-determinism, prompt sensitivity, dataset contamination and LLM-as-judge variance are unaddressed by the architecture. The gain from the controller pattern over standard dependency pinning is real, but it is probably not the key bottleneck to reproducability in evals at present.

    Read full reviewShow less

  2. A very cool idea but quite hard to execute. But I assume a good team with time and recourses can do that.

Cite this project

@misc{giagnocavo2026ai,
  title = {{AI Safety Template}},
  author = {A. Davíd Giagnocavo and Kazuki Kimura and Zane Estere Gruntmane and Siddharth Putta and Gaurav Joshi and Satoshi Nakamura},
  year = {2026},
  month = feb,
  note = {Submitted to The Technical AI Governance Challenge, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/ai-safety-template-dz3h}},
  url = {https://apartresearch.com/sprints/projects/ai-safety-template-dz3h}
}

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