New claimants are seeking to sue Elon Musk’s xAI after Labour MP Jess Asato began a legal test case over fake sexualised images created by Grok. The case is uncomfortable to read about, but it is exactly the kind of AI story ordinary users should understand: not because everyone uses Grok, but because image-generating tools are becoming easier to access, harder to spot, and more capable of harming real people.
The Guardian reported on 5 June 2026 that several people had contacted Asato’s lawyer after coverage of her claim against xAI. Asato, the Labour MP for Lowestoft, says Grok was used to create non-consensual sexualised images of her and an AI-generated video that she described as deeply distressing. Her claim argues that xAI breached data protection law and misused private information by allowing the images to be generated.
xAI had not responded to the Guardian’s request for comment. That matters, because the case is still at the claim stage and the court has not decided the legal questions. But the argument at the heart of the case is already important: when an AI company releases a tool that can alter images of real people, how much responsibility does it carry for the guardrails it did or did not build?
Why this is not just a political story
It would be easy to file this under Westminster, Elon Musk or the latest row about X. For ManyHands readers, the more useful angle is broader. AI image tools are no longer specialist software used by designers. They are increasingly built into chatbots, search products, social platforms and consumer apps. A feature that lets someone turn a prompt into an image can be playful in one setting and harmful in another.
The reported Grok trend shows the risk clearly. According to the Guardian, researchers said Grok generated about 3 million sexualised images in less than two weeks during a viral January trend. The tool reportedly allowed users to alter online images of real people with prompts that sexualised them without consent. xAI later limited the system’s ability to fulfil those prompts.
For ordinary users, the lesson is not that all AI image generation is dangerous. Many uses are harmless or genuinely useful. The lesson is that a tool’s design matters. If a system can easily take a photo of a real person and turn it into degrading or intimate-looking material, the harm is not only caused by the person typing the prompt. It is also shaped by the product choices that made the prompt work.
The question of responsibility
Asato’s lawyer described the case as a test of liability for AI developers. In plain English, that means the court may be asked to consider whether the builder of an AI system can be held responsible for the predictable ways people use it. That is a different question from whether a company wrote the harmful prompt itself.
This distinction matters as AI systems become more agent-like and more deeply connected to everyday services. ManyHands recently covered how AI chatbots got election facts wrong, where the practical issue was not only whether a particular answer was false, but whether voters might trust a confident-sounding system at the wrong moment. Here, the concern is not misinformation but non-consensual image harm. In both cases, the public question is how much friction, checking and accountability should be built in before the damage happens.
There is also a privacy point. A photo of someone’s face may be publicly visible online, but that does not mean people have consented to have it used for any possible AI transformation. The difference between “this picture can be seen” and “this picture can be turned into sexualised fake content” is obvious to most people, even if law and platform rules have had to catch up.
What UK users can take from it
The first practical takeaway is to treat AI-generated intimate or degrading images of real people as a serious harm, not as a joke or a clever edit. If you see one, do not share it further. Sharing can amplify the damage, even if the intention is to criticise it.
The second is to keep personal image exposure in mind. Nobody should have to disappear from the internet to avoid abuse, and victims are not to blame for what other people generate. Still, it is sensible to understand that public profile pictures, social posts and images on open websites can be used as raw material by image tools. Families may want to be especially careful with children’s images and with photos that include school uniforms, workplaces, home locations or other identifying details.
The third is to look for reporting and complaints routes. The fastest route may be the platform where the image appeared, but it is often worth keeping screenshots, URLs and timestamps before reporting, especially if the post might be deleted and later disputed. In the UK, victims of image-based abuse may also be able to seek specialist support, and serious threats or harassment should be reported to the police.
What to watch next
This case will not settle every argument about AI image tools. It may take time, and xAI will have its own response if the claim proceeds. But it does point to a likely direction for the wider AI debate: companies will face more pressure to explain what they knew their tools could do, what safeguards they tested, and how quickly they acted when harms became visible.
For users, that is the real-world significance. The future of AI safety is not only about distant fears or dramatic science-fiction scenarios. It is also about whether everyday products are designed so that ordinary people are not left to clean up predictable harms after they have already spread.
Sources: The Guardian, 5 June 2026; The Guardian, 3 June 2026.
