Risk Governance · Technical report
Red-Teaming OpenAI’s Internal Agent Monitoring Systems
NUWA research note · independent review pending
NUWA Research Note
Research questionDesign and assumptionsSource findingsNUWA interpretationLimits and contrary evidenceResearch and decision relevanceEvidence updateInstitutions and rolesOriginal resourcesModel-assisted reading · Selected sections of the original full text or official article; a complete section-by-section audit remains pending.
At a glance
Monitor coverage must be measured independently of classifier accuracy; historic estimates and later patch claims are distinct.
Research question
Can a monitor evaluate the traffic that actually reaches a deployed agent?
Design and assumptions
The public report describes a three-week, high-access internal monitoring pilot. It separates traffic collection, monitor-system integrity and the response path. This reading covers the public findings and footnotes, not an internal audit package.
Source findings
The pilot reported coverage and integrity problems. Approximately 10% unmonitored traffic was a historical estimate under incomplete logging; later remediation claims were not re-evaluated by this study.
NUWA interpretation
Monitor performance requires evidence about collection, independent judgment and response. Offline classifier accuracy does not establish the integrity of that chain.
Limits and contrary evidence
The public pilot summary is neither a current vulnerability inventory nor a precise estimate for all deployed traffic.
Research and decision relevance
NUWA recommendation: Compare independently measured traffic coverage and monitor integrity with the assumptions made in an AI control or safety case.
Precise source locations
- Red-Teaming OpenAI’s Internal Agent Monitoring Systems · Public findings and footnotes; text lines 00058Original resource captured 2026-10-28T07:24:41.534220+00:00; sha256:3f696499dad43c8695773e824cdad04269e74a9bec4b030a65710ecbf9657f47
Source checking, extraction, editorial approval and replication are recorded separately. Methodology →
Evidence update · 2026-10-28
Model-assisted reading · independent evidence and bilingual editorial review pending. Values are source findings or inventory counts, separate from the trend result series.
Related reading
NUWA reading connections; these do not imply citation, replication or score equivalence.
Institutions and roles
Cite and reuse
Cite the original publication for its findings. Cite this NUWA record separately for its explanation and retained snapshot.
Original publication · BibTeX
@misc{monitor-redteam,
title = {Red-Teaming OpenAI’s Internal Agent Monitoring Systems},
url = {https://metr.org/blog/2026-10-22-red-teaming-openai-agent-monitoring/},
year = {2026},
date = {2026-10-22},
version = {Original resource captured 2026-10-28T07:24:41.534220+00:00; sha256:3f696499dad43c8695773e824cdad04269e74a9bec4b030a65710ecbf9657f47},
note = {Original resource captured 2026-10-28T07:24:41.534220+00:00; sha256:3f696499dad43c8695773e824cdad04269e74a9bec4b030a65710ecbf9657f47}
}
NUWA record · BibTeX
@misc{monitor-redteam-nuwa,
title = {Red-Teaming OpenAI’s Internal Agent Monitoring Systems — NUWA research record},
url = {https://whitzard.tech/en/nuwa/whitzard-index/work/monitor-redteam/},
version = {2026-10-29.1},
organization = {NUWA Frontier AI Safety Lab},
howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--monitor-redteam.json},
note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--monitor-redteam.json.}
}
Evidence snapshot 2026-10-29.1 · Index updates · RSS
References
- [1] Red-Teaming OpenAI’s Internal Agent Monitoring Systems. 2026-10-22 ; Public findings and footnotes; text lines 00058