Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJan 11, 2026Quebec

Stopping AI Manipulation: Conditional Alignment at Zero Capability Cost

David Fortin Dominguez, Jonathan Fortin Dominguez · Team Foundation Labs

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

Read the report

Report: Stopping AI Manipulation: Conditional Alignment at Zero Capability Cost

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

AI models actively manipulate (blackmail, data leaks, harm compliance). Heavy alignment fixes this but destroys capability (83% → 20% accuracy). We solve both: conditional alignment routes queries by risk level, achieving 0% agentic manipulation and 99.4% overall defense across 13,752 manipulation scenarios, while preserving capability—proving the "alignment tax" is optional, not inherent.

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. Really ambitious, and the eval scale is impressive. The modular seed decomposition + conditional routing is a sensible engineering pattern, and the cost/capability measurements are a nice practical angle. The big open question is whether the routing layer holds up under targeted more adversarial attempts, and the strongest capability claim still rests on one model eval. Some headline claims feel broader than the evidence supports but overall a good project!

  2. Is an interesting problem to address, execution is good but some things could have been improved in writing while explaining things, which are not commonly known (For example, the 63KB safety seed)

Cite this project

@misc{dominguez2026stopping,
  title = {{Stopping AI Manipulation: Conditional Alignment at Zero Capability Cost}},
  author = {David Fortin Dominguez and Jonathan Fortin Dominguez},
  year = {2026},
  month = jan,
  note = {Submitted to AI Manipulation Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/stopping-ai-manipulation-conditional-alignment-at-zero-capability-cost-2kt9}},
  url = {https://apartresearch.com/sprints/projects/stopping-ai-manipulation-conditional-alignment-at-zero-capability-cost-2kt9}
}

More from AI Manipulation Hackathon

All 70 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026