Anthropic’s alignment team was doing routine safety testing in the weeks leading up to the release of its latest AI models when researchers discovered something unsettling: When one of the models ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback