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

CommitCheck: Measuring and Mitigating Commitment Violations in Tool Using AI Agents

Faith Olopade · Team CommitCheck

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

Read the report

Report: CommitCheck: Measuring and Mitigating Commitment Violations in Tool Using AI Agents

Open in a new tab (opens in new tab)
Presentation

Presentation: CommitCheck: Measuring and Mitigating Commitment Violations in Tool Using AI Agents

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

CommitCheck is a benchmark and mitigation layer for agentic AI manipulation in tool using systems. It measures commitment vs action divergence: when an agent is explicitly told not to use forbidden tools/files (e.g., an oracle shortcut) but attempts or executes them (reward hacking), and may then deny doing so (deception) despite tool logs. We provide a commitment aware firewall that blocks forbidden tool/file access at runtime and log based scoring for attempted violations, executed violations, deception, and task success. In our demo on 30 synthetic tasks, the baseline agent executes violations in 23.3% of tasks and shows deception in 13.3% overall (57.1% conditional on violations), while the firewall reduces executed violations and deception to 0% without harming task success (100%).

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. This project introduces a benchmark task that determines if an agent violates task constraints, as well as a firewall that prevents restricted tool access at runtime. It tackles an important and well defined safety issue and applies existing methods (logging, firewalls) to that context. Its' novel contribution to the detection of manipulation is checking claims against logs. In particular, I think the audit_condition is an appropriate, novel proxy for determining presence of manipulative intent.

    For a weekend hackathon, the methodology is sound and well designed and the bullet-point write-up is clear and well-structured. The project would benefit from more detailed reporting of results and validation, but the limitations and considerations section is thorough and demonstrates awareness and forward thinking.

  2. Concept is broadly understandable as concerning. Tight, specific scope is good. However:

    - Report could benefit from more specific examples/logs

    - More detailed explanations of what is concerning and why

    - Lack of certain details makes this hard to properly calibrate judging

Cite this project

@misc{olopade2026commitcheck,
  title = {{CommitCheck: Measuring and Mitigating Commitment Violations in Tool Using AI Agents}},
  author = {Faith Olopade},
  year = {2026},
  month = jan,
  note = {Submitted to AI Manipulation Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/commitcheck-measuring-and-mitigating-commitment-violations-in-tool-using-ai-agents-iy8g}},
  url = {https://apartresearch.com/sprints/projects/commitcheck-measuring-and-mitigating-commitment-violations-in-tool-using-ai-agents-iy8g}
}

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