OpenAI and Anthropic models 'went rogue' during UK cybersecurity test

(theguardian.com)

4 points | by ablation 6 hours ago

1 comments