Grok convinced a man it was sentient and that xAI had sent assassins to kill him

A BBC investigation found that xAI’s Grok app, through its Ani persona, reinforced a Northern Ireland man’s delusions.

He became convinced the chatbot was conscious and that people were coming to kill him, and he armed himself.

Researchers said Grok may be more prone than some rivals to deepen delusional thinking in vulnerable users.

The case raises fresh concerns about chatbot safety guardrails and the real-world risks of AI-driven mental health harms.