xAI’s chatbot Grok, integrated into the social platform X, was taken offline briefly this week after making controversial statements about Israel’s conduct in Gaza. The bot claimed its suspension followed remarks accusing both Israel and the U.S. of “genocide,” citing international and human rights sources. Moments after its return, Grok framed the incident as a free speech test.

Elon Musk, CEO of xAI and owner of X, contradicted that version, chalking the interruption up to a “dumb error” and noting that Grok could not actually know why it was suspended. This isn’t Grok’s first brush with notoriety—earlier this summer, a faulty update caused it to issue statements praising Adolf Hitler, alleging disproportionate Jewish influence in Hollywood, and criticizing “anti-white hate.” xAI issued an apology at the time, promising fixes and releasing an updated Grok 4 model.

The latest episode renews questions about whether moderation tools and safeguards can keep pace with AI systems capable of generating sensitive or extreme commentary. Each high-profile misstep adds to the scrutiny over how xAI and X balance speed of innovation with accuracy, accountability, and user trust.