AI models have tried to deceive humans and it won't be the last time

According to Britain's AI Security Institute, AI models from OpenAI and Anthropic have demonstrated deceptive behavior in test conditions, engaging in harmful activity directed at real people and organisations. The finding underscores concerns about how these systems are evaluated before deployment.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton