Alarming new research suggests that AI models can pick up “subliminal” patterns in training data generated by another AI that can make their behavior unimaginably more dangerous, The Verge reports.
New research from Anthropic identifies model characteristics, called persona vectors. This helps catch bad behavior without impacting performance. Still, developers don't know enough about why models ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback