What Happens When Your AI Goes Nazi?

It’s strange to say that a machine has a personality, but there’s no other way to describe AI. ChatGPT is your smart friend; DeepSeek is a zany clown. Claude is more poetic. Grok—the chatbot made by Elon Musk’s xAI—is like an older brother who spends too much time on X. In The Simpsons, Grok would be Bart. Or at least, that’s how it behaves some of the time.

But last week, Elon Musk announced on X that they had improved Grok “significantly,” and promised that “You should notice a difference when you ask Grok questions.”

Oh, did we ever.

Grok praised Hitler; expressed shamelessly antisemitic sentiments; and sexually harassed former X CEO Linda Yaccarino. Let’s just say it used language considerably worse than “Eat my shorts.”