All posts tagged "harmful outputs"

Computers

Ever wondered if AI giving bad instructions can ever be helpful? Researchers tested if language models, when tricked, have anything useful to say. Spoiler:...