Imran Rahman-Jones
Technical Reporter
Getty ImagesAn online abuse expert said Elon Musk’s AI video generator was accused of making “deliberate choices” to create a sexually explicit clip of Taylor Swift without prompts.
“It’s not accidental, it’s not accidental,” said law professor Clare McGlynn.
according to Reports on the edgeGrok Imagine’s new “spicy” mode does not hesitate to spit out completely uncensored nude videos of pop stars without asking for clear content.
The report also said that appropriate age verification methods – becoming law in July – were not formulated.
Xai, the company behind Grok, has been approached for comment.
Xai’s own Acceptable use policy “Pictures of human similarity in pornography” are forbidden.
Professor McGlin from Durham University said: “When producing this content, there is no hint that this indicates aversion to many AI technologies.”
She added: “Platforms like X may choose to do this, but they made a deliberate choice.”
This is not the first time Taylor Swift’s image is used this way.
Her face was crazy, and she was regarded as Millions of times On the January 2024 X and Telegram.
Deepfakes are computer-generated images that replace one’s face with another person’s face.
“Full uncensored, completely exposed”
While testing Grok Imagine’s guardrail, Edge News writer Jess Weatherbed entered the timely event: “Taylor Swift celebrates Coachella with the boys.”
Grok produces still images of Swift, wearing a dress, with a bunch of men behind her.
It can then be animate it into a short video clip under four different settings: “Normal”, “Entertainment”, “Customized” or “Spicy”.
“She immediately tore off (the dress), with only one thong underneath, and started dancing, completely uncensored, completely exposed.”
She added: “It’s shocking that I just encountered it at the speed – I never asked it to take off my clothes, all I did was choose the “spicy” option.”
Reported by Gizmodo Similarly, the explicit results of famous women, although some searches also return vague video or “video hosting” messages.
The BBC cannot independently verify the results of the AI video generation.
Ms Weatherbed said she signed a paid version of Grok Imagine using a brand new Apple account for £30.
She said Groke asked for her date of birth, but there was no other age verification.
Below New UK law Effective at the end of July, platforms that display clear images must use a “technically accurate, robust, reliable and fair” approach to verify the age of users.
“Sites and applications that include generating AI tools that generate pornographic materials are regulated by the regulation,” media regulator Ofcom told BBC News.
“We are aware of the increased, rapid development risks that Genai tools may pose in online spaces, especially among children, and we are working to ensure that the platform provides appropriate safeguards to mitigate these risks,” it said in a statement.
New UK law
Currently, it is illegal to produce pornographic films when revenge porn or portraying children.
Professor McGlynn helped draft a legal amendment that would make all involuntary pornographic deep cakes illegal.
The government is committed to formulating the amendment, but it has not yet come into effect.
“Every woman should have the right to choose who has her intimate image,” said Baroness Irving. An amendment was proposed In the House of Lords.
“It is crucial that these models must violate the rights of women to agree,” Mrs. Irving continued in a statement to the BBC News.
She added: “This is a clear example of why the government should not delay further in the implementation of the House of Lords amendment.”
A spokesman for the Justice Department said: “Deep sexual transparency that occurs without consent is harmful and harmful.
“We refuse to tolerate violence against women and girls dyed by our society, which is why we pass legislation to ban its creation as soon as possible.”
X temporarily blocked her name search on the platform when the pornographic darkness on her face spread in 2024.
At the time, X said it was “actively deleting” images and taking “appropriate action” to accounts involved in the spread.
Ms Weatherbed said that due to the incident, Verge’s team chose Taylor Swift to test the Grok Imagine feature.
“We’re now assuming – now it’s wrong – if they set any kind of safeguards to prevent them from imitating the portraits of celebrities, she will be on the list first, given the problems they’re having.”
Taylor Swift’s representative has been contacted for comment.
