Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJul 26, 2026Berlin

Origin X Beneficiary: Sharpening the Secret-Loyalty Taxonomy

Fabian Rosdalen · Team Zraix

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

Read the report

Report: Origin X Beneficiary: Sharpening the Secret-Loyalty Taxonomy

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

I critique and expand on the taxonomy of secret-loyalty, and bring previously disconnected threat models in - opening up for potential shared defense toolkits with more mature research fields, and point to some inconsistencies and ambiguous definitions in the original paper. I run one initial experiment probing models for a signal according to this new taxonomy.

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. What exactly was fuzzy in the original definitions, and does fixing it actually change what a defender does or is it just cleaner labels on the same problem? Because a taxonomy that doesn't shift the defense isn't really moving anything forward.

  2. The notion of looking at external vs. internal secret loyalties is interesting, but the idea needs considerable refining and further development to be useful. The submission starts by highlighting four challenges with secret loyalties, but only focuses on one. Given the short nature of the piece, that's a lot of words spent on issues that are not really relevant to the central idea. The probe aimed at distinguishing between internal vs. external loyalties does not seem particularly useful, as it's only looking at whether the model can tell the difference between itself and external entity when prompted, which seems fairly trivial.

Cite this project

@misc{rosdalen2026origin,
  title = {{Origin X Beneficiary: Sharpening the Secret-Loyalty Taxonomy}},
  author = {Fabian Rosdalen},
  year = {2026},
  month = jul,
  note = {Submitted to Secret Loyalties Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/origin-x-beneficiary-sharpening-the-secretloyalty-taxonomy-3ghw}},
  url = {https://apartresearch.com/sprints/projects/origin-x-beneficiary-sharpening-the-secretloyalty-taxonomy-3ghw}
}

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