Anthropic opens privacy-safe Claude usage data pilot to independent researchers
On August 26, Anthropic announced completion of a pilot program providing independent access to aggregate Claude usage data via Anthropic Insights, its privacy-preserving analysis tool. Stanford SALT Lab, Oxford Human Information Processing Lab, and AI safety nonprofit METR each designed their own studies analyzing approximately 250,000 Claude conversations.
On August 26, Anthropic announced completion of a pilot program providing independent access to Anthropic Insights, its privacy-preserving analysis tool. Stanford's SALT Lab, Oxford's Human Information Processing Lab, and AI safety nonprofit METR each designed their own studies analyzing approximately 250,000 Claude conversations from April-May 2026.
Key findings include that over half of Claude conversations involved consequential tasks, and newer Claude models deliver significant productivity speedups. Anthropic is now accepting expressions of interest from researchers for future participation.
Anthropic Insights is Anthropic's privacy-preserving analysis tool designed to allow external researchers to study AI usage patterns while protecting user privacy. This is an important step in Anthropic's push for transparency in AI safety research.