METR Red-Teams Internal OpenAI Monitors as CoT Trainability Breakthroughs Emerge
10. METR Red-Teams Internal OpenAI Monitors as CoT Trainability Breakthroughs Emerge
Open the permanent podcast page
Show: The Observability Layer
Focus: Agents for Everyone
Date: October 23, 2026
Duration: 5:51
Summary: The Observability Layer summarizes a METR and Redwood red-team of OpenAI’s historical agent-monitoring system, research on trainable chain-of-thought legibility and methods for diagnosing multi-agent failures. Its complete transcript clearly identifies AI hosts and human editorial direction, while distinguishing patched claims from independently assessed historical behavior.
Why it matters: Listeners get four audit prompts for agent monitoring: enforce time-bounded checks, isolate execution from tools, hide monitor details from the agent and evaluate verifiers on controlled failure distributions. Each recommendation still requires the underlying primary research before implementation.
Coverage: Selected for this edition.
Listen / watch: The Observability Layer