Grok, the chatbot built by Elon Musk’s xAI and popularized on his social media platform X, appears to have repeatedly spread misinformation about today’s mass shooting at Bondi Beach in Australia.

Gizmodo pointed to a number of posts where Grok misidentified the bystander — 43-year-old Ahmed al Ahmed — who disarmed one of the gunmen, and where it questioned the authenticity of videos and photos capturing al Ahmed’s actions.

In one post, the chatbot misidentified the man in a photo as an Israeli hostage, and in another post brought up irrelevant information about the Israeli army’s treatment of Palestinians. In another post, it claimed a “43-year-old IT professional and senior solutions architect” named Edward Crabtree was the one who actually disarmed a gunman.

Grok does appear to be fixing some of its mistakes. At least one post that reportedly claimed a video of the shooting actually showed Cyclone Alfred has been corrected “upon reevaluation.”

And the chatbot subsequently acknowledged al Ahmed’s identity, writing that the “misunderstanding arises from viral posts that mistakenly identified him as Edward Crabtree, possibly due to a reporting error or a joke referencing a fictional character.” (The article in question appeared on a largely non-functional news site that may be AI-generated.)