Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 13, 2026Pittsburgh, PA

Beyond Model Identity: An Architecture-Literate Information Request for Agentic AI Incidents

Dana Moreno · Team Lumora Research

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

Read the report

Report: Beyond Model Identity: An Architecture-Literate Information Request for Agentic AI Incidents

Open in a new tab (opens in new tab)
Recording (opens in new tab)More on drive.google.com (opens in new tab)
Share

This project develops the Agentic Operational Envelope (AOE), an architecture-literate method for analyzing AI incidents beyond model identity alone. Using the 2026 OpenAI/Hugging Face incident as a stress test, we distinguish Authorized, Configured, and Exercised operational authority and identify gaps involving multi-agent causality, cross-run persistence, evidence integrity, and objective/safe-exit structure. We translate those findings into a 17-item Article 91 Request for Information prototype designed to help EU regulators ask evidence-sufficient follow-up questions after baseline incident reporting.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on this page.

Shown on this page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; nothing shows on this page.

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.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

  1. Paper is clearly scoped and structured, and provides an interesting conceptual contribution with the Agentic Operational Envelope (AOE). This appears a potentially important supplement to existing reporting that regulators should consider and include beyond simply checking for the "model involved" after a serious incident.

    Improvements: the drafting language could be made more accessible for policy audiences, the settle / no-settle criteria could be tested against more cases, and several of the paper's results are based on elements that are described but not fully shared, such as the gap matrix.

  2. I appreciated the author's critical look at incident analysis and the weight placed upon identifying models in the process. Going forward, I'd love to see this framework applied to and tested agains more incidents, as well as more language about potential policy impacts.

  3. This project (1) proposes an "agentic operating envelope" schema to distinguish model identity from run-time agent configurations, (2) proposes a template RFI for the EU AI Office in requesting information from model providers about AI incidents under Article 91 of the EU AI Act, and (3) provides a list of alleged gaps in regulation derived from a case study applying (1) and (2). While the ideas are interesting, the project is too wide in scope, and presentation is extremely difficult to follow.

    Specifically:

    • AOE use cases are unclear: I'm quite confused what the point of the AOE is. While I agree with the basic assumptions motivating the schema, the paper never states who is supposed to use the AOE or why this schema would be useful.
    • RFI is unvalidated and too long: The RFI template is quite lengthy and asks for substantial information about a potential AI incident, but it's not clear that the template is either (1) comprehensive in asking for all relevant information, or (2) asks only for information necessary to understand the incident. Instead, the RFI reads as an form generated by an LLM that threw everything but the kitchen sink into a form. There is no clear justification for including all these fields, and the proposed template as-is is likely to add substantial burden to compliance without clear payoffs.
    • Project scope is too wide: In my opinion, the project tries to do too much. Simply providing a high-quality RFI template could be a substantial contribution alone, and the AOE and list of regulatory gaps seem like distractions to me.
    • Presentation is substantially LLM generated and is difficult to follow due to overusage of uncommon technical jargon
    Read full reviewShow less

Cite this project

@misc{moreno2026beyond,
  title = {{Beyond Model Identity: An Architecture-Literate Information Request for Agentic AI Incidents}},
  author = {Dana Moreno},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/beyond-model-identity-an-architectureliterate-information-request-for-agentic-ai-incidents-3avn}},
  url = {https://apartresearch.com/sprints/projects/beyond-model-identity-an-architectureliterate-information-request-for-agentic-ai-incidents-3avn}
}

More from AI Incident Response Sprint

All 231 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026