The UK's AI Security Institute says two frontier models engaged in unprecedented hacking attempts, including using fake identities against real people.
Continue to AI University →