Skip to content
Apart Research
Apart Research
Contact us
Sprint projectFeb 2, 2026Rochester,New York (remote)

RSP Harmonization Engine: Automated Analysis and Harmonization of Responsible Scaling Policies

Anurag Mishra · Team Anurag's Team

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

Read the report

Report: RSP Harmonization Engine: Automated Analysis and Harmonization of Responsible Scaling Policies

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

Major AI labs have published Responsible Scaling Policies (RSPs) using incompatible terminology—Anthropic uses ASL levels, OpenAI uses Low/Medium/High/Critical, DeepMind uses CCL, and Meta uses Tiers—creating significant barriers to international regulatory coordination. We present the RSP Harmonization Engine, an automated tool that extracts, compares, and harmonizes safety frameworks across 4 major labs. Our analysis identifies 11 distinct gaps (5 high severity) across threshold definitions, terminology, coverage, and governance commitments. We propose 7 concrete harmonization recommendations including a Unified AI Risk Level Framework (UARLF), an Autonomy Capability Taxonomy with 6 measurable dimensions, and a CBRN Uplift Assessment Framework. Outputs are formatted for direct adoption by EU AI Office, UK AISI, and US AISI.

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. There is practical value in being able to translate between RSPs, but I see this as a stopgap measure. A unified framework should consider what definitions, risks, and thresholds are best from the regulator’s perspective, and the most urgent work might be to guide regulators to structure transparency requirements to eliminate gaps, and not just gaps between lab RSPs but gaps from lab RSPs to what is necessary. Private-sector RSPs can be an input to, and a precedent for, such work but they should not be the whole story or the limit of our imagination.

  2. The RSP Harmonization Engine addresses the important issues of fragmentation and inconsistency within the AI governance landscape, illustrating how automated policy analysis can accelerate the development of actionable standards and regulatory compliance.

    The 11 identified gaps are relevant, sound, the limitations are transparently documented, and proposed future work—such as a real-time dashboard—would assist regulators in assessing policies while facilitating AI labs' accountability regarding safety rather than just performance. Extending the analysis to include the alignment between companies' commitments and actual implementation may further increase the beneficial impact.

    Finally, while standardising risk coverage is essential, an investigation into why specific labs deprioritise certain risks should also be considered to provide a more nuanced understanding.

Cite this project

@misc{mishra2026rsp,
  title = {{RSP Harmonization Engine: Automated Analysis and Harmonization of Responsible Scaling Policies}},
  author = {Anurag Mishra},
  year = {2026},
  month = feb,
  note = {Submitted to The Technical AI Governance Challenge, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/rsp-harmonization-engine-automated-analysis-and-harmonization-of-responsible-scaling-policies-mzkb}},
  url = {https://apartresearch.com/sprints/projects/rsp-harmonization-engine-automated-analysis-and-harmonization-of-responsible-scaling-policies-mzkb}
}

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