OpenAI research finds models get harder to monitor via CoT as intelligence scales
tszzl · x · 2026-11-12
New OpenAI research from jeffreygwang examines how CoT monitorability (no-CoT capability and CoT controllability) evolves over the model training lifecycle. The findings are "unsurprising but sobering": as systems scale toward greater intelligence, models become harder to monitor through their chain-of-thought. kaicathyc amplified the thread, urging much stronger work on alternatives to CoT monitoring and on making monitorability itself a training objective.
Related event: OpenAI Study Finds Stronger Models Have Less Monitorable Chain-of-Thought(3 posts)→
More from Safety
- UK elite schools ditch laptops while poor pupils get AI chatbot tutors — AlexTensor · 2026-11-13
- Ought launches Epistemics Fund with $500k–$5M for Community Notes research — jungofthewon · 2026-11-13
- Claude nearly executes instructions read from an article, sparking injection concerns — npmer · 2026-11-13
- TSMC confirms leak of ~46K employee records offered for 3 BTC after cloud vendor breach — rohanpaul_ai · 2026-11-13
- TSMC hacked: Nina Schick argues security is downstream of intelligence — NinaDSchick · 2026-11-13
- 22% of Nvidia sales went to Singapore, fueling China smuggling claims — SumitGup · 2026-11-13