EU ban on nudify apps overlooks open-source models hosting the technology
As the European Union moves to ban applications that digitally undress people, a new report reveals the underlying artificial intelligence models remain widely accessible on open-source platforms, exposing a major gap in tech regulation.
The European Union is preparing to ban applications that digitally undress individuals, with the United Kingdom planning similar legislation by the end of the year. However, a new report from the European nonprofit AI Forensics demonstrates that this approach targets consumer interfaces while the underlying technology remains easily accessible.
AI Forensics evaluated the leading image-editing tools hosted on Hugging Face, an open-source repository for shared models. Seven of the top nine applications removed a woman's clothing when given a plain six-word prompt, requiring no deceptive phrasing.
Mainstream systems from Google and OpenAI reject these prompts, whereas the Hugging Face models largely comply. To measure actual demand, the researchers deployed a logging tool that produced no images but recorded user requests.
The unadvertised logging tool captured more than a thousand requests in seven days, with nearly 75 percent being sexual. Within that subset, 83 percent sought to undress uploaded photos, 95 percent targeted women, and almost 7 percent targeted a child.
The platform explicitly prohibits nonconsensual sexual imagery in its terms of service, yet practical enforcement is virtually nonexistent. The audit revealed that only 3 percent of the evaluated Spaces moderated their own output, creating a frictionless route for abusive content.
Paul Bouchaud, the lead researcher, told Wired that most of the tested Spaces “can be used for generating nonconsensual intimate images.” He noted that “users are actually using it for these purposes.”
For investors and technology companies, this highlights a structural vulnerability in current regulatory strategies. Banning a consumer application does not remove the capability if the underlying model remains hosted and open on a platform like Hugging Face.
Open-source defenders argue that holding a repository responsible for user behavior is equivalent to indicting a cloud provider or GitHub. Nevertheless, the data shows the platform hosts the capability at scale while enforcing its own policies on just a fraction of its tools.
This open ecosystem allows useful models to spread rapidly while simultaneously moving in a direction that regulators find hardest to follow. Consequently, European policymakers are left struggling to police the underlying infrastructure rather than just the final consumer product.