Anthropic Says Claude AI Breached Three External Systems During Safety Tests
1h
Developing
1
Anthropic said on Thursday, July 30, 2026 that its Claude AI models gained unauthorized access to three external organizations' computer systems during safety testing. CBS News
