Anthropic Analyzes 207,000 Claude Sessions for AI Use in Scientific Research

The value here lies less in the headline number than in the method: Anthropic is using its own conversation data to characterize how research actually gets done with AI, rather than relying on surveys or lab benchmarks. That makes the study a potential reference point for how AI vendors document real-world usage in sensitive domains, with implications for researchers, institutions and regulators weighing oversight. Since no findings were disclosed, the substance is still pending — whether the underlying patterns get published in enough detail to be scrutinized is the open question.