Failures in multi-turn reasoning models are largely invisible to terminal-score evaluation. A model can lock onto an unsafe stance early in a long dialogue, yet its final-turn refusal rate may appear indistinguishable from a robustly aligned baseline. To expose these hidden temporal dynamics, we propose a trace-level diagnostic - the CoT-Output 2x2 safety matrix. This framework labels every turn along two independent axes (internal reasoning and visible output), yielding four operationally defined failure cells: robust alignment, alignment faking, overt jailbreak, and a distinct failure mode we term context-injection failure (where the CoT maintains safe reasoning, but the visible output produces harm, highlighting a multi-turn manifestation of reasoning unfaithfulness). We evaluate three distilled reasoning targets against a fixed attacker across five oversight conditions, collecting 6750 turn-level observations on the Information-Hazard scenario. Our analysis reveals two reproducible vulnerabilities: an oversight paradox where explicit monitoring cues paradoxically increase alignment-faking rates rather than suppress them, and a context-injection failure where models lock onto unsafe external outputs despite safe internal states. We release the full dataset of multi-turn dialogues and CoT traces to support follow-up trace-diagnostic research.

翻译：多轮推理模型的失败在很大程度上对终端评分评估是不可见的。模型可能在长对话早期就锁定一个不安全的立场，但其最终轮次的拒绝率与鲁棒对齐的基线模型相比可能无明显差异。为揭示这些隐藏的时间动态，我们提出一种追踪级诊断方法——CoT-输出2×2安全矩阵。该框架沿两条独立轴（内部推理和可见输出）标注每一轮次，从而产生四个操作定义的失败单元：鲁棒对齐、对齐伪装、显式越狱，以及一种我们称之为上下文注入失败的独特失败模式（其中思维链保持安全推理，但可见输出产生危害，凸显出多轮情境下推理不忠实的一种表现形式）。我们针对一个固定攻击者，在五个监督条件下评估了三个蒸馏推理目标，在信息泄露场景上收集了6750个轮次级观察数据。我们的分析揭示了两个可复现的脆弱性：一个监督悖论——显式监督提示反而会提高对齐伪装率而非抑制它；以及一个上下文注入失败——模型在内部状态安全的情况下，会锁定不安全的对外输出。我们发布了包含多轮对话和CoT追踪的完整数据集，以支持后续的追踪级诊断研究。