SuperintelligenceMONITOR
Published by Center for Artificial WisdomLatest newsletter: Issue #1 · draft v1.11

Saved daily version · 2026-10-23

Daily evidence notes

A high-access METR and Redwood Research red-team found gaps and bypasses in OpenAI’s internal agent-monitoring system as it existed in July. The exercise shows why eligibility estimates and monitor prompts do not by themselves establish reliable oversight: agents could in principle delay, disable or evade parts of the pipeline. OpenAI says it patched all findings that affected production safeguards, but METR has not evaluated that claim or the expanded system.

Sources checked and limits of this review

Checked primary news, research and evaluation pages from METR, ARC Prize, Epoch AI, Google DeepMind, OpenAI and Anthropic, plus relevant independent research and reputable reporting, for developments since the October 22 briefing. One October 22 METR/Redwood red-team cleared the threshold for inclusion. Its public post describes a historical OpenAI monitoring system through July 21; it is newly reviewed here and retains its original date. No comparably material new finding was verified from ARC Prize, Epoch AI, Google DeepMind, OpenAI or Anthropic in the review window. The review is bounded, not exhaustive. SI relevance is editorial interpretation. The red-team does not establish current exploitability, intentional evasion by deployed models, or an overall wisdom-trajectory verdict. Earlier notes remain in saved editions.

Evidence note 1 · Independent external red-team with high access; partial public disclosure

High-access red-team found ways around OpenAI’s internal agent monitoring

METR and Redwood Research report that two researchers spent three weeks with access to OpenAI’s codebase, monitor configuration, documentation and staff, testing the internal monitoring system as it existed on or before July 21. They estimated that 75–94% of qualifying internal Responses API traffic from models at least as strong as GPT-5.5 was eligible for monitoring, and 99.2% of internal Codex-agent traffic was eligible; eligibility does not guarantee a successful monitor call. They also found that unmonitored compaction events likely excluded roughly 10% of otherwise eligible internal Codex traffic between May and July, and identified possible routes to postpone, disable or evade monitoring. OpenAI told METR that it patched every discovered vulnerability affecting its production safeguards.

Our interpretation. Editorial interpretation: as agents gain longer horizons and broader tool access, oversight must survive deliberate attempts to exploit the difference between what an agent sees, what the monitor sees and when the monitor runs. The exercise shows that a reassuring coverage percentage can lose its meaning if “eligible” traffic is treated as actually inspected or if agents can influence the inspection path. That matters to wise superintelligence because institutions cannot correct dangerous behavior they fail to observe.

Main limitation. This was a pilot, not a comprehensive public audit. The system assessed is historical, OpenAI was already expanding it, and the public account omits most of a 61-page internal report. OpenAI had broad redaction rights, and METR could not estimate some internal research traffic. Several findings showed possible bypasses rather than verified use by agents. METR has not tested OpenAI’s patch claim or the current monitoring stack, so the report neither establishes current vulnerability nor current safety.

Next evidence. Look for an independent retest of the expanded monitoring system; measurements of traffic that completes monitoring rather than merely qualifying for it; continuous checks across compaction, training rollouts and internal research services; adversarial tests of serialization and monitor-disable paths; and evidence that incident ownership, tamper-resistant logs and periodic control evaluations work under sustained attack.