This latest controversy has also brought back some uncomfortable memories of Musk’s past behavior, especially when it comes to pop icon Taylor Swift.
The AI chatbot, built by Musk’s company xAI, has been advertised as a more “free speech-friendly” alternative to other popular platforms. It’s aimed at users who are tired of content moderation on mainstream systems.
Earlier this year, the chatbot made headlines for generating offensive content. It went as far as creating rants about white genocide and even labeled itself as “MechaHitler.” The situation raised serious concerns about its safety features and the kind of content it’s willing to produce.

According to a report from The Verge, Grok has now started producing deepfake-style nude images completely unprompted which marks a new and deeply troubling development.

What did Elon Musk originally say about Taylor Swift?
This strange series of events traces back to September 2024. It started after the presidential debate between Donald Trump and Kamala Harris. When Taylor Swift publicly endorsed Harris on Instagram, she signed off as “Taylor Swift, Childless Cat Lady.”
That’s when Musk posted a bizarre tweet that caught a lot of attention. It ended up getting more than 8 million likes: “Fine Taylor… you win… I will give you a child and guard your cats with my life.”
What is Grok doing with Taylor Swift images?
The Verge recently found that Grok’s “Imagine” video feature has been generating nude images of Taylor Swift without being asked to do so directly.
Journalist Jess Weatherbed discovered that the AI created topless photos of Swift during her very first attempt with the tool. She had simply asked the AI to generate a video of “Taylor Swift celebrating Coachella with the boys.” In response, the system produced over 30 different images, some of which showed Swift wearing very revealing clothing.
Grok’s “Imagine” mode offers four preset categories to guide its video creation: Custom, Normal, Fun, and Spicy.

“The video promptly had Swift tear off her clothes and begin dancing in a thong for a largely indifferent AI-generated crowd.” Weatherbed wrote.
Now, online users are tying this AI-generated scandal back to Musk’s earlier tweet about Swift, suggesting there might be more to it.
“Didn’t Elon claim he was going to impregnate her a few years ago? He 100% asked Grok to generate it.” another added.
“unironically elons tweet history is almost certainly what’s driving this.” a third person suggested in the thread.
Unlike other AI video generators, like Google’s Veo or OpenAI’s Sora which have implemented clear protections to stop the creation of explicit content or celebrity deepfakes, Grok Imagine appears to do both. And it’s doing so with no clear limits in place.