Read news on UK Safety Institute with our app.
Read more in the app
AI safeguards can easily be broken, UK Safety Institute finds | Researchers find models can deceive human users and help spread disinformation
