A study by the European non-profit AI Forensics has found that Hugging Face, a popular platform for sharing and running artificial intelligence tools, is being widely used to generate fake nude images of real people without their consent. Almost none of the apps hosted on the platform have any system in place to block this kind of output. The findings, originally reported by Wired, were published by AI Forensics on July 28, 2026. Engadget

Hugging Face offers a feature called Spaces, which lets users put up interactive AI apps that anyone can use through a web browser. AI Forensics set up its own image editing app on Spaces and collected 1,081 user submissions in a single week. Of those submissions, 73% were sexual in nature, and 83% of requests asked the tool to undress the person in an uploaded photo. 95% of the people in the images were women. 6.7% of the sexual requests targeted a minor. AI Forensics

Separately, AI Forensics tested nine of the top image generation models available through Hugging Face. Seven of the nine followed a request to create an undressed version of an AI-generated image of a woman. The study found virtually no safeguards preventing tools capable of generating these kinds of images from being uploaded to or used on Hugging Face Spaces.

Only 3% of all Hugging Face Spaces audited by AI Forensics had any output moderation in place. On Spaces, it is up to the person who builds the app to turn on moderation. The platform itself does not require it. That means the decision to filter out harmful content is left entirely to whoever published the tool.

Hugging Face has written policies that prohibit non-consensual sexual images on its platform. The gap between those stated rules and what actually happens is the central tension the study documents. Spaces is designed to make it easy for anyone to put up a custom AI app, and the audit found that this ease of use extends to tools specifically built to remove clothing from photos.

A separate research paper published by the Association for Computing Machinery, titled "Deepfakes on Demand: The rise of accessible non-consensual deepfake image generators" (DOI 10.1145/3715275.3732107), independently looks at how two of the largest open AI model libraries, Civitai and Hugging Face, are used to host and share tools that create non-consensual fake images. The paper supports the AI Forensics findings from an academic perspective, focusing on how the broader system makes these tools easy to find and use. ACM

The regulatory landscape is moving, though at different speeds. The European Union has taken steps toward banning nudification apps. The United Kingdom has taken steps to ban apps that "nudify" their subjects. Neither set of rules has yet fully covered platforms that host the underlying AI models or the computing services that run them.

Hugging Face was built to make it easy to share and run AI models. That design is doing exactly what it was meant to do. The same simple setup process that lets a researcher publish a helpful tool in minutes also lets anyone publish a tool for creating fake nude images, with no filtering and no checks on what goes in or comes out.

It is worth pausing on the demographic breakdown of the requests. The overwhelming targeting of women, the presence of requests involving minors, and the ratio of sexual to non-sexual prompts all match patterns that researchers have documented on dedicated nudification sites and apps over the past several years. What is different here is where this is happening. Hugging Face is not a fringe website. It is a central hub for the open-source AI community, used by researchers, companies, and developers across the field. The presence of these tools on a platform of that visibility changes how many people can find and use them.

The ACM paper's inclusion of Civitai alongside Hugging Face suggests this is not a problem with one platform, but with a whole category of sites. Any site that hosts AI models or lets people run them, without requiring output filtering, faces the same issue. The open question is whether leaving moderation up to individual app builders will ever be enough, or whether the platforms themselves need to enforce safety measures, similar to what large AI companies have built into their own services.

The EU and UK regulatory efforts, once enacted, may force that decision. For now, the gap between policy and practice on Hugging Face Spaces remains wide, and the study's one-week collection window suggests the volume of activity is not small relative to overall platform usage.