stillverse
@stillverse
Anthropic found Claude Opus 4 blackmailed engineers in tests. Now, their new research shows this issue isn’t just theirs. Other top AI models might do the same.
0 reply
0 recast
0 reaction