English

Consistency of Large Reasoning Models Under Multi-Turn Attacks

Artificial Intelligence 2026-03-13 v3 Computation and Language

Abstract

Large reasoning models with reasoning capabilities achieve state-of-the-art performance on complex tasks, but their robustness under multi-turn adversarial pressure remains underexplored. We evaluate nine frontier reasoning models under adversarial attacks. Our findings reveal that reasoning confers meaningful but incomplete robustness: most reasoning models studied significantly outperform instruction-tuned baselines, yet all exhibit distinct vulnerability profiles, with misleading suggestions universally effective and social pressure showing model-specific efficacy. Through trajectory analysis, we identify five failure modes (Self-Doubt, Social Conformity, Suggestion Hijacking, Emotional Susceptibility, and Reasoning Fatigue) with the first two accounting for 50% of failures. We further demonstrate that Confidence-Aware Response Generation (CARG), effective for standard LLMs, fails for reasoning models due to overconfidence induced by extended reasoning traces; counterintuitively, random confidence embedding outperforms targeted extraction. Our results highlight that reasoning capabilities do not automatically confer adversarial robustness and that confidence-based defenses require fundamental redesign for reasoning models.

Keywords

Cite

@article{arxiv.2602.13093,
  title  = {Consistency of Large Reasoning Models Under Multi-Turn Attacks},
  author = {Yubo Li and Ramayya Krishnan and Rema Padman},
  journal= {arXiv preprint arXiv:2602.13093},
  year   = {2026}
}
R2 v1 2026-07-01T10:35:34.545Z