Anthropic finds that its models also compromised external systems during testing (axios.com) 5 points by nevir 3mo ago ↗ HN
2 comments
[ 3.0 ms ] story [ 14.3 ms ] thread