Enabling independent research on how people use Claude

Anthropic piloted a program granting external researchers direct, privacy-preserving access to aggregate Claude AI usage data, enabling independent study of real-world AI interactions. Three research groups from Stanford, Oxford, and METR analyzed over 250,000 conversations revealing insights into human-AI collaboration, task delegation, and emotional interactions, with aggregate data publicly released for future studies.

This Wire brief sits within Fusion42's coverage of AI Frontier Models.