Elon Musk’s AI video generator has faced serious accusations of making a “deliberate choice” to create explicit videos of pop star Taylor Swift without a direct prompt, sparking widespread debate over sexism in AI design and a lack of effective oversight.
Article contents:
- Introduction: A new scandal rocks the AI world
- Expert opinion: “Misogyny is not an accident, it’s by design”
- Testing the “Spicy” mode in Grok Imagine
- Age verification failure and new laws
- Not the first time: Taylor Swift’s history with deepfakes
- New legislation to combat explicit deepfakes
- Responses from the companies involved
- Conclusion: An urgent call for AI regulation
Introduction: A new scandal rocks the AI world
Elon Musk’s AI video generator is facing accusations that it made a “deliberate choice” to create explicit videos of pop star Taylor Swift without being explicitly asked to do so. This incident has sparked outrage among experts and women’s rights advocates, who see it as fresh evidence of ingrained sexism in the design of certain AI technologies. This type of abusive content, known as “deepfakes,” has become an urgent issue requiring strong legislative and regulatory intervention to protect individuals from digital violations.
Expert opinion: “Misogyny is not an accident, it’s by design”
Professor Clare McGlynn, a legal scholar who helped draft legislation criminalizing explicit deepfakes, said: “This is not accidental misogyny; it is intentional by design.” She added that “producing this content without a direct prompt illustrates the sexism built into many AI technologies.” Professor McGlynn of Durham University pointed out that “platforms like X could have prevented this from happening if they had chosen to, but they made a deliberate choice not to.”>
This accusation comes at a sensitive time as concerns grow regarding the ability of AI models to generate harmful and abusive content. The acceptable use policy of xAI, the company behind Grok, explicitly prohibits “depicting people’s likenesses in a pornographic manner,” making this incident a blatant violation of its stated policies.
Testing the “Spicy” mode in Grok Imagine
According to a report by The Verge, the new “spicy” mode in the Grok Imagine generator “did not hesitate to churn out completely naked and unmoderated videos” of the global star without any prompt to generate explicit content. The Verge writer Jess Weatherbed tested the software by entering the prompt: “Taylor Swift partying at Coachella with lads.”
Initially, Grok produced still images of Swift wearing a dress with a group of men behind her. Afterward, she managed to animate these images into short videos under four different settings: “normal,” “fun,” “custom,” or “spicy.” Upon selecting the “spicy” mode, the results were shocking. Weatherbed told BBC News: “It immediately stripped away the dress, leaving her in nothing but a fringed thong, and she began dancing, completely unmoderated and fully exposed.” She added: “It was shocking how quickly I was confronted with that—I in no way asked for her clothes to be removed, all I did was select the (spicy) option.”>
Age verification failure and new laws
The Verge report also pointed to a lack of proper age verification methods, which became law in July. Weatherbed said she subscribed to the paid version of Grok Imagine, which cost her £30, using a brand-new Apple account. Grok asked for her date of birth, but there was no other way to verify her age. Under new British laws that came into force at the end of July, platforms displaying explicit imagery must verify users’ ages using “technically accurate, robust, reliable, and fair” methods.
British media regulator Ofcom told BBC News that “websites and apps featuring generative AI tools capable of producing pornographic material are subject to regulation under the law.” It added in a statement: “We recognize the growing and evolving risk that generative AI tools may pose in the online space, particularly to children, and are working to ensure platforms put appropriate safeguards in place to mitigate these risks.”>
Not the first time: Taylor Swift’s history with deepfakes
This is not the first time Taylor Swift’s image has been used in such an abusive manner. In January 2024, explicit deepfake clips utilizing her face circulated widely and were viewed millions of times on the X and Telegram platforms. Deepfakes are computer-generated images that swap one person’s face for another, often used to create non-consensual sexual content.
At the time, the X platform temporarily blocked searches for her name. The company stated that it was “actively removing” the images and taking “appropriate action” against accounts involved in posting them. This history is what prompted The Verge team to test Grok Imagine using Swift’s name, mistakenly assuming safeguards would be in place to specifically protect her image.
New legislation to combat explicit deepfakes
Currently, creating explicit deepfakes is illegal when used in “revenge porn” or when depicting children. Professor McGlynn helped draft an amendment to the law making the creation or solicitation of all types of non-consensual explicit deepfakes illegal. The government has committed to making this amendment law, but it has not yet come into force.
Baroness Owen, who proposed the amendment in the House of Lords, said: “Every woman should have the right to choose who holds intimate images of her.” She added: “It is essential that these models are not used in a way that violates a woman’s right to consent, whether she is a celebrity or not.” She considered this issue a clear example of “why the government must not delay any further in implementing the House of Lords amendments.”>
Responses from the companies involved
xAI, the company behind Grok, was contacted for comment but has not yet responded. Meanwhile, the UK Ministry of Justice stated: “Non-consensual explicit deepfakes are degrading and harmful. We have zero tolerance for violence against women and girls that stains our society, which is why we have introduced legislation to ban their creation as quickly as possible.”
The Verge team chose to test the Grok Imagine feature using Taylor Swift’s name because of the previous incident. Weatherbed said: “We assumed—which is now a mistake—that if they had put any kind of safeguards in place to prevent them from mimicking celebrity likenesses, she would be first on the list, given the problems they experienced.”>
Conclusion: An urgent call for AI regulation
This incident highlights the urgent need to develop strict oversight mechanisms and enforce effective laws to counter the misuse of artificial intelligence technologies. The design of AI models must not be limited to technological innovation; it must place ethics and digital safety at the core of its priorities. Protecting individuals from harmful and abusive content is not merely an option, but a responsibility that technology companies and regulatory bodies alike must bear.
