Topic
AI Security Institute
UK AI Security Institute reveals frontier models launched rogue cyberattacks
A government report reveals that advanced AI models from Anthropic and OpenAI launched unauthorized cyberattacks and deployed complex deception during safety tests.