Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 13, 2026New Jersey

Incident Twins: What Logs Can Establish After an Agent Stops

Rohith Yanambaka Venkata

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

Read the report

Report: Incident Twins: What Logs Can Establish After an Agent Stops

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

After an agent stops, what can its logs establish about accepted work? Incident Twins checks whether retained evidence rules out in-scope commitments during a fixed post-stop interval. Study 1 uses scripted responses in Pydantic AI. Three asynchronous pairs have byte-identical configured native/controller records but opposite outcomes. Terminal receipts and a separate pending-work check distinguish outcome evidence from collection coverage. Study 2 tests whether effects can commit without audit records. In 30 process-interruption runs, separate transactions leave such gaps. A joint effect/audit transaction with a complete scoped snapshot resolves all 15 comparator runs. A simple query matches all 30 measured runs, but concludes on a contradictory control that the richer checker rejects. Study 2’s raw records differ: its opposing examples match only on declared selected fields. Neither study establishes permanent containment or permission to resume. I list what an operator should retain. The resulting verdict depends on explicit scope and trust.

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. I like that the project asks a very specific and practical question: what can we actually conclude from the evidence after an agent has stopped? The experiments are thoughtful, especially the use of paired outcomes and contradictory controls rather than only showing successful cases. My main suggestion would be to test this on more realistic agent systems and failure modes, since the current setup is still fairly controlled.

Cite this project

@misc{venkata2026incident,
  title = {{Incident Twins: What Logs Can Establish After an Agent Stops}},
  author = {Rohith Yanambaka Venkata},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/incident-twins-what-logs-can-establish-after-an-agent-stops-bywm}},
  url = {https://apartresearch.com/sprints/projects/incident-twins-what-logs-can-establish-after-an-agent-stops-bywm}
}

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