Grok

Chain-of-Thought Hijacking

Large Reasoning Models (LRMs) improve task performance through extended inference-time reasoning. Although previous studies suggest that lo…