Grok’s track record is spotty at best. But even by the very low standards of xAI, its failure in the aftermath of the tragic mass shooting at Bondi Beach in Australia is shocking. The AI chatbot has repeatedly misidentified 43-year-old Ahmed al Ahmed, the man who heroically disarmed one of the shooters, and claimed the verified video of his deed was something else entirely — including that it was an old viral video of a man climbing a tree.

In the aftermath of the attack, Ahmed has been widely praised for his heroism, but some have tried to dismiss or even deny his actions. Someone even quickly whipped up a fake news site that appears to be AI-generated, with an article naming a fictitious IT professional, Edward Crabtree, as the man who disarmed the attacker. This, of course, got picked up by Grok and regurgitated on X.

But Grok also suggested that images of Ahmed were of an Israeli being held hostage by Hamas. And it claimed that video taken at the scene was actually of Currumbin Beach, Australia, during Cyclone Alfred.

Broadly, it seems Grok is having trouble understanding queries today and working out the proper answers. It replied to a question about Oracle’s financial difficulties with a summary of the shooting at Bondi Beach. When asked about the validity of a story about a UK police operation, it first stated today’s date, then coughed up poll numbers for Kamala Harris.