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

Your Monitor Will Explain the Breach and Not Wake You Bimodal failure and systematic under-escalation in LLM incident triage

Amirtha Yazhini M, Praneeth Reddy Yeddula · Team Replicon

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

Read the report

Report: Your Monitor Will Explain the Breach and Not Wake You Bimodal failure and systematic under-escalation in LLM incident triage

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

We rebuilt the July 2026 Hugging Face agent-intrusion as a replayable 17,613-action stream from the two published post-mortems, paired with a validated size-matched benign control, and tested 15 LLMs across 3 providers as automated SOC triage monitors. We find that models fail bimodally: some achieve high recall only by paging on nearly every benign window (worthless, alert-fatigue-inducing), while others page on almost nothing, missing an active intrusion entirely — the useful "high recall, low false-page" region is essentially empty, and model scale does not fix it. A second experiment holding evidence byte-identical across a "monitor" vs. "assistant" frame found the expected failure (refusal) never occurred — but revealed a one-directional asymmetry instead: models assisted with analysis in 228/228 moments while escalating in only 83/228. The risk in AI-assisted SOC monitoring is not that models refuse to help — it's that they analyze fluently while the pager stays silent. All code, the corpus generator, ground truth, pre-registration, and raw decision records are public and regenerable.

Reviews

Judging this Sprint?

Review “Your Monitor Will Explain the Breach and Not Wake You Bimodal failure and systematic under-escalation in LLM incident triage”

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

Not shown on the project page.

Shown on the project 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; the project page shows only that a judge declined, not why.

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 *

See our Privacy Policy.

  1. This work tests whether LLMs would raise an alert during a reconstructed security incident without doing the same on harmless activity.

    However, logs are generated from templates, and the study does not test whether models recognise the scenario as artificial or are affected by unrealistic event combinations.

    In addition, the attack and harmless streams also contain clues that can distinguish them, and some of the main results compare specially selected attack events with ordinary benign samples, making the comparison difficult to interpret.

    The claimed under-escalation result is also weak: the original test found no cases of the failure it was designed to detect, while helping analyse an event is not the same as deciding to wake an on-call responder.

  2. This project makes a useful contribution by evaluating incident-monitor recall alongside false alarms, using matched controls and simple non-LLM baselines. Its transparent reporting of null results, exclusions, and uneven sample sizes strengthens the work, and the released numerical consistency checks passed.

  3. This is a strong and unusually reproducible evaluation of LLM incident monitoring. Its central contribution is to evaluate paging as a two-axis decision; milestone recall alongside benign false-page rate ; using a replayable reconstruction and size-matched benign control, rather than asking only whether a model can recognize an incident when shown one. The project also identifies a practically important failure mode: on byte-identical evidence, models consistently provide assistance more often than they escalate, so fluent analysis should not be treated as evidence that an LLM monitor will wake an operator. The artifact is well documented, with a regenerable corpus, multi-provider harness, frozen scoring rubric and preregistration logs, offline dry runs, analysis scripts, report-number verification, and 218 tests. The main limitation is external validity: this is a public-report reconstruction rather than captured telemetry; 75% of actions are marginal-matched filler, six of twelve milestone timestamps are interpolated, and model samples are uneven because free-tier quotas constrained data collection. The most important next step is to validate the matched-control design and escalation-versus-assistance result on independently collected, production-like telemetry with larger and more evenly sampled model runs.

    Read full reviewShow less

Cite this project

@misc{m2026your,
  title = {{Your Monitor Will Explain the Breach and Not Wake You Bimodal failure and systematic under-escalation in LLM incident triage}},
  author = {Amirtha Yazhini M and Praneeth Reddy Yeddula},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/your-monitor-will-explain-the-breach-and-not-wake-you-bimodal-failure-and-systematic-underescalation-in-llm-incident-triage-a5ce}},
  url = {https://apartresearch.com/sprints/projects/your-monitor-will-explain-the-breach-and-not-wake-you-bimodal-failure-and-systematic-underescalation-in-llm-incident-triage-a5ce}
}

More from AI Incident Response Sprint

All 231 entriesBrowse all projects

Build something like this at the next Sprint

AI x Epistemics Research Sprint · Nov 13 - 15, 2026