AI RESEARCH
Chain-of-Thought Hijacking
arXiv CS.AI
•
ArXi:2510.26418v4 Announce Type: replace Large Reasoning Models (LRMs) improve task performance through extended inference-time reasoning. Although previous studies suggest that longer reasoning should lead to robust safety behavior, we find evidence to the contrary: over-extended reasoning can instead be exploited to systematically weaken refusal behavior. We propose Chain-of-Thought Hijacking, a simple yet effective black-box jailbreak attack that induces LRMs to engage in prolonged benign puzzle-solving reasoning, often lasting than five minutes, before eliciting harmful compliance.