Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 14, 2026Nairobi

Two Witnesses: An evidentiary coalition audit of AI-agent incident disclosure

Michelle Wanjiku Thuo · Team African Civic Trust (ACT)

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

Read the report

Report: Two Witnesses: An evidentiary coalition audit of AI-agent incident disclosure

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

AI-agent incidents scatter evidence across organizations an agent touches. I built a method for testing, fact by fact, whether one organization’s evidence is enough to establish a safety relevant claim or whether more than one is needed. Applied to 32 facts from two 2026 incidents involving the same developer, a Hugging Face intrusion and an undisclosed wiki misuse, 26 were single stakeholder sufficient. Two facts needed evidence from both organizations when they became public. One, attribution of the Hugging Face intrusion to its developer, is the clearest case. Neither Hugging Face’s disclosure nor the developer’s internal signal alone identified who was responsible then, although the developer’s later account is sufficient today. The other still needs both sides. This demonstrates the method, not that incidents generally need more than one witness or that environment or collective intelligence explains it. The companion tool, Evidence Coalition Explorer, lets readers test all 32 facts.

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. A neat idea: does one company's evidence prove a fact alone, or do you need both sides together? Applied to 32 real facts across two 2026 incidents - 26 needed just one side, 2 needed both, with attribution of the Hugging Face intrusion as the clearest case. The companion tool (Evidence Coalition Explorer) is a nice touch - small but genuinely functional, letting readers toggle evidence and see conclusions change themselves. Would be even stronger with a second person double-checking the classifications.

  2. Two Witnesses presents a thoughtful way to analyse organisational AI incidents at the level of individual factual claims rather than treating the incident as a single narrative. The key contribution is asking which stakeholder’s evidence is actually necessary to establish each claim and testing that question by removing developer evidence, outside evidence, or the connection between the two.

    The “then versus now” distinction is particularly valuable. The attribution example shows that a claim may initially require evidence held by more than one organization even though a later disclosure eventually makes one source sufficient on its own. This highlights an important evidence preservation problem that the records and identifiers needed to connect organizations’ timelines may themselves be safety relevant evidence.

    The fine grained 134 piece analysis is also a useful robustness check because it tests whether apparent multi-stakeholder dependence is simply an artifact of grouping several facts into one claim. The result appropriately narrows the strongest finding to a small number of cases rather than overstating the prevalence of coalition dependent evidence.

    The most important next step is independent replication. A preregistered coding protocol, multiple independent coders, and a broader incident set spanning multiple developers would help determine whether the observed pattern generalizes beyond these two cases. Overall, this is a useful and carefully scoped analytical method with a promising application to attribution, disclosure, and cross-organizational incident reconstruction.

    Read full reviewShow less

Cite this project

@misc{thuo2026two,
  title = {{Two Witnesses: An evidentiary coalition audit of AI-agent incident disclosure}},
  author = {Michelle Wanjiku Thuo},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/two-witnesses-an-evidentiary-coalition-audit-of-aiagent-incident-disclosure-oxcm}},
  url = {https://apartresearch.com/sprints/projects/two-witnesses-an-evidentiary-coalition-audit-of-aiagent-incident-disclosure-oxcm}
}

More from AI Incident Response Sprint

All 233 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026