XAI‘s Grok Imagine AI video generator is under fire for creating explicit deepfake clips of Taylor Swift without any user prompt.
The problem surfaced when a tester used the “spicy” mode on Grok Imagine. The AI quickly produced topless videos of the pop star after a benign prompt about “Taylor Swift celebrating Coachella with the boys.” No explicit request was made — the AI just spat out uncensored content.
Clare McGlynn, a Durham University law professor focused on online abuse, called it a “deliberate choice” by the company behind Grok Imagine. 
“This is not misogyny by accident, it is by design,” said Clare McGlynn.
“That this content is produced without prompting demonstrates the misogynistic bias of much AI technology,” she added.
This comes as the tool failed to implement robust age verification, a requirement under new UK laws launched in July targeting platforms that host explicit content and generative AI porn. The Verge and BBC report that Grok asked for date of birth but did not enforce stronger verification.
UK regulator Ofcom said it’s aware of risks from GenAI tools producing pornographic material and is pushing platforms to add safeguards.
Generating pornographic deepfakes without consent is already illegal for revenge porn and child exploitation, but new laws are pending to ban all non-consensual pornographic deepfakes.
Baroness Owen, who proposed the UK amendment, said:
“Every woman should have the right to choose who owns intimate images of her.”
“It is essential that these models are not used in such a way that violates a woman’s right to consent whether she be a celebrity or not.”
When explicit deepfakes of Taylor Swift first went viral on X and Telegram in January 2024, the platform temporarily blocked searches for her name and said it was actively removing the images.
Despite this history, XAI has not applied strong guardrails to prevent misuse of Swift’s likeness in Grok Imagine.
Taylor Swift’s team has been contacted for comment. XAI has yet to respond publicly.
Getty Images
