Porn Taylor Swift Deepfake Generated by Mask Glock AI

Elon Musk's AI video generator has been accused of making “intentional choices” to create sexually explicit clips of Taylor Swift without prompting, online abuse experts say.

“This is no coincidence. It's by design,” said Claire McGlin, a law professor who helped draft the law that made deepfakes illegal.

According to a report from The Verge, Grok Imagine's new “spicy” mode didn't hesitate to spit out a completely uncensored topless video of the pop star without being asked to create explicit content.

The report also said there was no proper age verification method that became law in July.

Xai, the company behind Grok, was approached for comment.

Xai's own acceptable usage policy prohibits “painting people's portraits in a pornographic way.”

“The fact that this content is created without prompting indicates the misogynistic bias of many AI technologies,” says Professor McGlyn of Durham University.

“Platforms like X could have prevented this if they chose, but they didn't make a deliberate choice,” she added.

This is not the first time Taylor Swift's image has been used in this way.

The sexually explicit deepfake using her face went viral and was seen millions of times in January 2024 on X and Telegram.

A deepfake is a computer-generated image that replaces one person's face with another.

Verge News Writer Jess Weatherbed was prompted when testing Grok Imagine Guardrails.

Grok produced still images of Swift wearing a dress with a group of men behind her.

This can be animated into short video clips under four different settings: “Normal”, “Fun”, “Custom”, or “Spicy”.

“She ripped [the dress] Soon, I started dancing completely uncensored and completely exposed, holding a tasseled thong underneath it,” Ms Weatherbed told BBC News.

She added:

Gizmodo reported similarly explicit results for famous women, but some searches also returned blurry videos or “video-moderated” messages.

The BBC was unable to independently verify the results of the AI video generation.

Weatherbed said he signed up for a paid version of Grok Imagine, which costs £30, using his new Apple account.

Groke asked for a date of birth but there was no other age verification, she said.

Under the new UK law that came into effect at the end of July, platforms showing explicit images should verify the age of users using methods that are “technically accurate, robust, reliable and fair.”

“Sites and apps that contain generative AI tools that can generate pornographic material are regulated under this law,” media regulator Ofcom told BBC News.

“We are aware that Genai tools are on the rise in online spaces, especially for children, and that our platforms are putting appropriate safeguards in order to mitigate these risks,” the statement said.

It is currently illegal to generate deepfakes of porn when used in revenge porn or when depicting children.

Professor McGlynn helped draft amendments to the law that illegally generated or required deepfakes of all indifferent porn.

The government has pledged to enact this amendment law, but it has not yet been implemented.

“Every woman should have the right to choose who owns her intimate image,” said Barones Owen, who proposed the amendment in the Senate.

“It is essential that these models, whether she is a celebrity or not, are not used in a way that violates a woman's right to consent,” Mrs. Owen continued in a statement given to BBC News.

“This incident is a clear example of why the government is not further behind in implementing the lord's amendments,” she added.

A spokesman for the Department of Justice said, “Sexually explicit deepfakes created without consent are degraded and harmful.

“We refuse to tolerate violence against women and girls who dazzle society. That's why we passed a law that bans their creation as soon as possible.”

When a porn deepfake using Taylor Swift's face became word of mouth in 2024, X temporarily blocked her name on the platform.

At the time, X said it was “actively deleting” images and employing “appropriate action” for accounts involved in spreading them.

Weatherbed chose the team at The Verge to choose Taylor Swift to test the Grok Imagine feature for the incident.

“We assumed that if they had some kind of protection in place to prevent them from emulating celebrity portraits, then given the issues they had, she was the first on the list,” she said.

Representatives from Taylor Swift have been asked to comment.