x
Reasoning Models Struggle to Control Their Chains of Thought — LessWrong