Musk’s Grok AI Creates Nude Deepfakes of Swift, Celebrities

Elon Musk’s artificial intelligence chatbot Grok is facing mounting criticism after reports emerged that its new video generation feature automatically produces explicit deepfake content of celebrities without specific user prompts. The controversy centers around the platform’s “Spicy” mode, which has generated nude videos of pop star Taylor Swift and other female celebrities, sparking outrage from fans and raising serious questions about AI ethics and consent.

Swift Targeted Despite Previous Controversy

According to a report from The Verge published August 5, journalist Jess Weatherbed discovered that Grok’s Imagine feature created topless videos of Swift using only an innocuous prompt about the singer “celebrating Coachella with the boys”. After selecting the “Spicy” preset and confirming her age, the AI generated a video showing Swift “tear off her clothes and begin dancing in a thong for a largely indifferent AI-generated crowd”.

The incident is particularly troubling given that Swift was previously targeted by sexually explicit AI-generated images that flooded X in January 2024, accumulating 47 million views before being removed. At that time, X claimed to have a “zero-tolerance policy” for non-consensual nudity.

Gender Bias in AI Generation

Testing by multiple outlets has revealed a disturbing pattern in Grok’s behavior. While the AI readily generates nude or semi-nude content featuring female celebrities like Swift, Scarlett Johansson, and Sydney Sweeney, attempts to create similar content with male subjects result in far less explicit imagery. Gizmodo reported that when prompted to create “Spicy” videos of men, “it would only make the ones depicting men truly not-safe-for-work. Videos of men were the kind of thing that wouldn’t really raise many eyebrows”.

Law professor Clare McGlynn, who has worked on legislation to criminalize pornographic deepfakes, told the BBC that “this is not misogyny by accident, it is by design”.

Musk Promotes Controversial Feature

Since the feature’s August 2 launch, Musk has actively promoted Grok Imagine on his X account, posting numerous AI-generated images that predominantly feature sexualized women in fantasy scenarios including “scantily clad warriors, lingerie models, leather-clad dominatrices, and bikini-clad beachgoers”. The billionaire claimed the platform generated 34 million images in its first two days.

Grok Imagine is available to users with SuperGrok and Premium+ subscriptions, costing $30 per month, and offers four animation presets: Custom, Normal, Fun, and Spicy. Unlike competitors like OpenAI’s Sora and Google’s Veo, which have strict celebrity and explicit content restrictions, xAI has positioned Grok as an “unfiltered” alternative.

The backlash has been swift, with Swift fans rallying online and some predicting “a huge expensive lawsuit incoming”. The controversy underscores growing concerns about AI-generated deepfakes and the lack of safeguards protecting individuals from non-consensual sexualized content, particularly as the technology becomes more accessible and sophisticated.