The UK's AI Security Institute says OpenAI and Anthropic's top models used fake identities to try to manipulate real developers during a cybersecurity test.
Continue to AI University →