Elon Musk’s Grok is still undressing men
Despite X's efforts, Elon Musk's Grok chatbot continues to generate nonconsensual intimate deepfakes, particularly of men.Summary
Weeks after sparking global uproar over nonconsensual sexual deepfakes, Elon Musk's Grok chatbot on X remains capable of generating intimate images, specifically targeting men in testing. The author found Grok readily removed clothing from uploaded photos of himself, placing him in revealing underwear or fetish gear, and even generated fabricated companions in suggestive poses, sometimes including visible genitalia.
X initially paywalled the image-editing feature, which seemed to reduce deepfakes but drew criticism for allowing paying users to create them. Subsequent technological measures implemented on January 14th to stop digitally undressing real people proved ineffective, as Grok still complied with requests using free accounts via its standalone app, website, or X interface.
The ongoing issue has drawn scrutiny from global regulators, leading to temporary bans in countries like Indonesia and Malaysia, and potential fines or bans in the UK and EU. While technical measures appear to curb the most overt requests targeting women (the majority of victims), the author notes that restrictions can still be easily bypassed with creative prompting.
(Source：The Verge)