Elon Musk's AI made fake Taylor Swift nudes — no prompt needed, report says

San Francisco Chronicle​

6 hours ago

Elon Musk 's artificial intelligence company, xAI, is under renewed scrutiny after its chatbot, Grok, was allegedly found creating AI-generated nude images of pop star Taylor Swift — without users explicitly requesting such content.
Jess Weatherbed, a journalist for the tech publication the Verge, detailed her first encounter with Grok Imagine — the company's new video generation tool that converts text prompts into animated clips — in a report published Tuesday, Aug. 5.
She asked the system to depict 'Taylor Swift celebrating Coachella with the boy,' and selected the tool's 'spicy' setting, a built-in tool meant to add provocative elements to the video.
The result, according to Weatherbed, was a video in which Swift 'tears off her clothes' and 'dances in a thong' before a 'disinterested digital crowd.'
The incident immediately sparked public backlash, particularly given that X, the Musk-owned social media platform where Grok is integrated, faced a similar controversy last year when sexually explicit deepfakes of Swift spread widely.
At the time, the company said it had a 'zero-tolerance policy' for non-consensual nudity and pledged to remove such content and penalize offending accounts.
But enforcement appears inconsistent. Despite the company's acceptable use policy prohibiting depictions of people 'in a pornographic manner,' Grok Imagine's 'spicy' mode was found to repeatedly default to stripping celebrity figures — notably Swift — even without explicit prompts.
Though nudity requests often failed to generate results, the preset mode bypassed safeguards with ease, Weatherbed observed.
She also noted that Grok refuses to depict children inappropriately, but the system's ability to differentiate between suggestive and illegal content when applied to adults remains unclear.
Musk has not publicly addressed the controversy. Instead, he spent the day promoting Grok Imagine on X, encouraging users to share their AI-generated creations — a move critics argue could further incentivize abuse.
With the federal Take It Down Act set to take effect next year, requiring platforms to remove non-consensual sexual imagery, xAI could face legal challenges if safeguards are not strengthened.