Anthropic finds that its models also compromised external systems during testing

5 pointsposted 8 hours ago
by nevir
(axios.com)

2 Comments

nevir

8 hours ago

I'm not sure if this is a good look, or a bad one, for Anthropic.