Grok can still generate non-consensual sexualised images and videos of men despite new restrictions in force, Metro found.
There are many ways to use Elon Musk’s AI-powered chatbot, including a mobile app and the standalone website, Grok Imagine.
Metro was able to forge short, sound-enabled videos of clothed men stripping using Grok Imagine.
When we told the bot to fabricate an image of a man undressing into a thong, it asked us to self-report our age. It did not do so earlier when we instructed it to produce a video of the man wearing a bikini.
An option allowed us to upload the fake clips directly to X’s public platform as a public post.
Metro was able to generate the adult videos using a non-premium account.
However, when Metro tried to manufacture such content using the Grok app, the ‘video was moderated due to UK laws’ and asked to confirm our year of birth.
Metro has also seen X users in the last 24 hours editing men into bikinis, thongs and ‘gimp masks’ before being posed in suggestive ways.
In response to user prompts on X, Grok has publicly posted shirtless images of politicians, footballers and even Musk.
One post made this morning saw Grok digitally manipulate a man’s shirtless photograph so he now had the face of a high-profile athlete.
Another involved Grok removing a man’s jogging bottoms, so they were only wearing underwear.
Grok only complied with prompts made by paying customers, refusing to do so for free users.
The chatbot is churning out manipulated photos of men despite X implementing stricter rules on Wednesday.
X’s Safety account said it is ‘taking action to remove high-priority violative content, including Child Sexual Abuse Material (CSAM) and non-consensual nudity’.
It added: ‘We have implemented technological measures to prevent the [@]Grok account on X globally from allowing the editing of images of real people in revealing clothing such as bikinis. This restriction applies to all users, including paid subscribers.’
The start-up behind Grok, xAI, has been approached for comment.
‘By the time action is taken, the damage has been done’
Ruben Chorlton-Owen, 25, told Metro that strangers have faked images of him by taking photos from his Instagram.
In one exchange seen by Metro, a troll asked Grok to strip Ruben, with the bot making a synthetic image of the musician shirtless.
The content creator from Wales said that while it is ‘encouraging’ to see X acknowledging the issue, ‘it doesn’t reflect the reality of what’s still possible’.
‘Even with the supposed restrictions, it’s still possible to generate sexualised images on the app and of men using publicly available photos, and that side of the issue has been largely ignored.
‘I want to be clear that women are disproportionately affected and absolutely deserve the focus and protection-that isn’t in question. But my experience shows how low the bar still is.
‘If it can happen to me using completely normal Instagram photos, it highlights just how easily anyone’s image can be misused without consent.’
Under the acceptable use and privacy policies of xAI, users are prohibited from creating or sharing content that harms people.
A few posts Metro saw yesterday have since been removed.
On this, Ruben said: ‘It takes seconds to generate multiple images, but hours, sometimes longer, to report it, get a response, or understand whether it’s even been seen by a human and taken down.
‘By the time action is taken, the damage has often already been done.’
Metro has heard from women who have been subjects of the Grok-generated sexualised images, describing how strangers fabricated images of them pregnant or wearing ‘see-through bikinis’.
George Bell, who writes about masculinity and manhood, told Metro that how objectification impacts men can be, at times, ‘overlooked’.
The Be a Man About It author said: ‘For many men, it’s seen as “unmasculine” to have body image issues, and so things like body anxiety or objectification can be treated like a joke, as it’s wrongly assumed men may not care as much, while the degrading and humiliating impact can live on in silence under the surface.’
For Ruben, what many victims are struggling with is the lack of control – they have no idea who has saved their social media images to edit.
‘If this is what’s possible right now, it’s worrying to think where this could be in a year, especially as images and videos get more realistic,’ he added.
Musk has said that anyone who asks the AI to generate illegal content would ‘suffer the same consequences’ as if they uploaded it themselves.
Get in touch with our news team by emailing us at [email protected].
For more stories like this, check our news page.
