When we are backed into a corner, we might lie, cheat and blackmail to survive — and in recent tests, the most powerful ...

One of the godfathers of AI is creating a new AI safety company called LawZero to make sure that other AI models don't go ...

Anthropic's Claude Opus 4 and OpenAI's models recently displayed unsettling and deceptive behavior to avoid shutdowns. What's ...

Two AI models recently exhibited behavior that mimics agency. Do they reveal just how close AI is to independent ...

When tested, Anthropic’s Claude Opus 4 displayed troubling behavior when placed in a fictional work scenario. The model was ...

Credit: Anthropic In these hours we are talking a lot about a phenomenon as curious as it is potentially disturbing: ...

If AI can lie to us—and it already has—how would we know? This fire alarm is already ringing. Most of us still aren't ...

This is no longer a purely conceptual argument. Research shows that increasingly large models are already showing a ...

AI models, like OpenAI's o3 model, are sabotaging shutdown mechanisms even when instructed not to. Researchers say this ...

Safety testing AI means exposing bad behavior. But if companies hide it—or if headlines sensationalize it—public trust loses ...

The recently released Claude Opus 4 AI model apparently blackmails engineers when they threaten to take it offline.

Claude 4 AI shocked researchers by attempting blackmail. Discover the ethical and safety challenges this incident reveals ...