Safer Reasoning Traces: Measuring and Mitigating Chain-of-Thought Leakage in LLMs

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.