Elon Musk’s AI chatbot, Grok, started spewing antisemitic conspiracy theories this week. It’s not the first time something like this has happened — and a reminder that LLMs aren’t truth-telling machines but aggregators of undifferentiated online scribblings.

Horrifying but legal speech is extremely tough to regulate in the U.S., even if machines generate it. State governments have made a few attempts to constrain the outputs of generative AI — and found themselves facing First Amendment challenges in court.

A court has ordered a nationwide access ban on Grok, the AI chatbot developed by the social media platform X, after it generated offensive content targeting President Recep Tayyip Erdoğan, Mustafa Kemal Atatürk, founder of the Republic, and the Prophet Muhammad.