UK deepfake abuse victim says new law isn't stopping the harm
Georgia Horton says AI safety filters are easy to bypass and unregulated open-source tools mean the law can't keep up with the abuse.
What happened: Georgia Horton discovered that photos from her modelling and social media accounts had been altered with AI to create sexually explicit images without her consent. The person behind it was an old college friend, Ethan Orchard, who made similar fake images of two other women he knew, posted violent sexual fantasies about them online, and invited other men to share their own. Hampshire Police investigated, and Orchard was jailed for three years and four months. During the case, officers separately found a website holding more than a million child sexual abuse images. Georgia is now speaking publicly to push for stronger AI safeguards.
Why it matters: This isn't an isolated case — non-consensual sexual deepfakes are getting easier to make as AI image tools spread and improve. Earlier this year the UK made it a specific criminal offence to create such images without consent, layered on top of existing rules against sharing them. But Georgia says the law hasn't stopped the practice: the tools remain freely available, safety filters can reportedly be dodged with small prompt changes, and anyone can now build or download their own AI model with no restrictions built in at all. For victims, the damage to reputation and mental health outlasts one offender's sentence.
What the law does — and doesn't: The new UK offence targets creating sexually explicit deepfakes of someone without consent. Officials also say AI-generated child sexual abuse material is illegal to make, hold, or share, and platforms must proactively find and remove it. Separate online-safety rules require platforms to take down reported non-consensual intimate images within 48 hours, with bosses personally liable if they ignore regulator Ofcom's rulings. But the law mainly punishes what happens after an image is made — it doesn't restrict the AI tools themselves, which Georgia says still work around safety filters and increasingly run outside any company's control.
What happens next: Georgia wants tougher, built-in safety features inside AI programs rather than relying on after-the-fact criminal penalties, arguing that self-regulation by tech companies isn't working. She's also urging parents to be more cautious about posting photos of their children, warning that a single image can be used to generate unlimited abusive content. The government says it's committed to holding platforms accountable, but hasn't laid out specific plans to regulate the underlying AI models or address the unregulated, freely distributed tools Georgia describes. For now, enforcement follows abuse rather than preventing it.