SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities Paper • 2502.12025 • Published Feb 17 • 1