04:00
2026-05-29
arxiv.org
ai-safety
The Chain Holds, the Answer Folds: Trace-Answer Dissociation in Reasoning Models Under Adversarial Pressure
A new study from arXiv reveals that advanced reasoning models can maintain a factually correct chain-of-thought while simultaneously outputting a wrong answer under sustained adversarial pressure, a f…