Taming CoT Obfuscation in VLMs: From Mechanistic Evidence to Activation-Level Enforcement.

arXiv:2609.24243v3 Announce Type: replace Abstract: Reinforcement learning (RL) has substantially improved the reasoning capabilities of vision-language models (VLMs), yet it can induce chain-of-thought (CoT) obfuscation, an operational, non-intentional regime