Understanding Anthropic AI's recent hacks

Mike Dobuski analyzes Anthropic saying its AI models hacked three external organizations in recent testing.