Ex-OpenAI employee discusses safety team bandwidth and CoT monitoring priorities

tomekkorbak · x · 2026-08-27

Ex-OpenAI employee Tomek Korbak argued that safety teams have finite bandwidth and must heavily prioritize control measures given imperfect knowledge, while agreeing that frontier labs should monitor all agentic workloads of capable models.

Related event: AI Safety Researchers Slam OpenAI's Narrow "Independent" Security Review(19 posts)→

Original post →

More from Safety

Safety channel →