Anthropic says Claude AI was used to build missiles, hunt Uyghurs and spy on 25 million phones: 5 'shocking cases' from its new threat report

Anthropic's new threat intelligence report details how its Claude AI was exploited between December 2025 and August 2026.