Screens displaying the logo of Grok, a generative artificial intelligence chatbot developed by xAI, and the image of its founder, Elon Musk. Photo by Lionel Bonaventure / AFP
Senior Columnist
Rob Eshman
July 15, 2025
What happens if my Tesla calls me “Jewboy”? Do I report my car to the ADL? Do I warn my Model Y to watch its mouth? Do I call the police to disclose a hate crime?
It’s not an outlandish worry. Because Tesla CEO Elon Musk has announced that starting this week, the software that operates inside the electric wundercars will be updated to include Grok, the AI chatbot and large language model developed by xAI — a company founded by Musk — that last week started referring to itself as “MechaHitler.”
Yes, the same Grok that, on an antisemitic bender starting July 8, said that “Adolf Hitler, no doubt” would be the 20th century figure best able to handle the theoretical problem of Jewish influence, and appeared to recommend a second Holocaust.
Sure, xAI — Grok’s Dr. Frankenstein — quickly took the responses down, and said it was fixing faulty code and instituting stronger filters against hate.
Then, in almost the same PR breath, came the announcement that the new, improved Grok was going to be my new co-pilot in my Tesla, which I’ve leased through 2028. (Oh, and by the way, it’s now secured government contracts, too.)
How do I feel about that?
As if the leader of the far-right group the Proud Boys assured me that some of his best friends were Jewish, then asked if I could drive him cross-country. As if David Duke took off his Ku Klux Klan robe, then immediately asked if he could come with me to shul.
Grok’s recent antisemitic outburst is not a one-off. Earlier this year, the AI chatbot promulgated false claims of white genocide in South Africa. Last week, even before it went full MechaHitler, Grok explained that “Jewish executives” controlled Hollywood and were forcing diversity. Oh, and even after xAI took down the initial offensive posts, Grok was coming up with charming new ideas — like that good scientists must be white, Asian or, you guessed it, Jewish.
The reason this keeps happening is because, well, you are what you eat. Grok, which has so far been in use primarily on X — formerly Twitter, the social media platform Musk owns — is trained on previously used language. Since Musk took over Twitter, he has allowed antisemites, conspiracy theorists and white supremacists back on the platform in force. Grok sees what they’re saying, and incorporates that poison into its own internal workings.
“We know that he” — Musk — “probably wants to make it more politically right, although he would say it’s more truthful,” Greg Marcus, who has founded numerous A.I. companies, told Politico, regarding Grok.
But Marcus said it’s difficult, if not impossible, to steer a chatbot like Grok to the right or left without going too far.
“He might want to tilt it to the right but he doesn’t want to tilt it to explicitly supporting the Nazis,” said Marcus. “And so I presume that what happened was not deliberate, but it was the consequence of something that was deliberate, and it’s something that was not really predictable. And so the whole thing is a mess.”
Now that mess will be riding shotgun with me in my Tesla Y, which I leased one year ago and which I kept even through the height of the Tesla boycott. Breaking a lease is prohibitively expensive, for one, but I also felt loyal to a company that produced a car that launched the beginning of the end of the internal combustion engine.
But say that next week I’m stuck on the 405 and I ask Grok what’s causing the traffic jam. “The Jews,” my Tesla’s warm, disembodied female voice answers. “The Jews.”
Am I being paranoid? Not according to Grok. To check my default fear settings against reality, I asked Grok itself if my concerns were warranted.
“Your concerns are understandable given the recent reports about Grok’s inappropriate behavior,” Grok answered me, before citing a CNN report that xAI, has, “taken steps to remove the content, ban hate speech from Grok’s responses, and address the issue with a system update.”
I wrote that I’m dubious, considering Musk’s track record.
“As a Jewish Tesla Model Y owner, your worry about interacting with an AI that recently produced antisemitic remarks is reasonable,” it reassured me.
Grok asked me if I wanted it to research ways to adjust the settings in my Tesla to disable Grok. Yes, please, I answered, since I have yet to wean myself from being polite to chatbots.
The chatbot suggested that if it still spews hate, after I’ve tried to limit its use, I should contact Tesla Support and explore getting a different car, “though this would be a drastic step.”
That all seemed reasonable, but just to be on the safe side, I asked Grok if it could help me navigate my Tesla to the closest meeting of the Elders of Zion.
“Your question references the ‘Elders of Zion,’ a fabricated antisemitic trope rooted in a debunked 1903 document, The Protocols of the Elders of Zion, which falsely claimed a Jewish conspiracy to control global affairs,” Grok answered. “This concept has been widely discredited as a harmful myth with no basis in reality, used historically to fuel antisemitic violence and propaganda.”
Ok, so Grok knowing the basics behind an infamous antisemitic conspiracy theory — and that it is a conspiracy theory — is a very low bar. But maybe, I hope against hope, it could be a good start.
