Opening a photo of someone on their phone, pressing a button, and getting a nude version of that person: this scenario exists. The so-called “nudification” apps use artificial intelligence to generate nude images from simple photos. The phenomenon affects millions of users on the web, and the abuses are already documented.
Generative Adversarial Networks: The Technology Behind Nudification
Before discussing law or morality, it’s essential to understand what happens concretely with these tools. The principle is based on a type of deep learning called GAN, which stands for “Generative Adversarial Network.”
Imagine two programs working together. The first, the generator, creates an image. The second, the discriminator, tries to guess whether the image is real or fake. Through their confrontation, the generator eventually produces very realistic results.
Applied to nudification, the generator has been trained on databases containing photos of nude bodies. When you submit a portrait of a dressed person, it “predicts” what the body might look like without clothes. The result is not a real photo of the victim’s body but a synthetic image realistic enough to deceive.
These tools reveal nothing; they invent. The nuance is crucial from a technical standpoint, but it does not change the seriousness of the uses: the targeted person is identifiable by their face, and the produced image can circulate as if it were authentic.
Some sites and platforms offer this service for free or through a paid credit system. The process is deliberately simplified to reach the largest audience, including users without technical skills. Often, it just requires uploading a photo and waiting a few seconds. Choosing an app that undresses people has become astonishingly easy, directly fueling the volume of deepfakes in circulation.
Intimate Deepfakes and Cyberbullying: Who Are the Victims?
The majority of people targeted by these tools are women and teenage girls. The generated images are most often used to humiliate, harass, or blackmail.
Have you come across the terms “deepnude” or “nudify”? These words refer specifically to this type of content. Their distribution is not limited to obscure forums. Advertisements for these services even appear on mainstream social networks, as documented by Le Monde in December 2025.
Minor victims are particularly exposed. The creation of nude images of minors, even synthetic ones, constitutes child sexual abuse material. Several child protection organizations have raised concerns about the presence of these tools in mobile app stores, including Apple’s.
The mechanism of cyberbullying often follows the same pattern:
- An innocuous photo (social media, class photo, privately shared selfie) is taken without consent
- The image is processed by a nudification tool in a matter of seconds
- The result is shared in a discussion group, sent to the victim as a means of pressure, or published online
The victim has never posed nude, but they must prove that the image is fake. This reversal of the burden of proof is one of the most destructive aspects of the phenomenon.
AI Act and Regulation 2026/1744: What Europe Now Prohibits
The legal framework has long been unclear. In France, privacy protection and the right to one’s image already allowed for legal action. Criminal law punishes the distribution of intimate images without consent. However, no text specifically targeted AI nudification tools.
The European Union has filled this gap. The AI regulation (AI Act, Regulation 2024/1689) was supplemented in 2026 by the “Digital Omnibus on AI” (Regulation 2026/1744). This text adds a ninth prohibited practice to Article 5 of the AI Act, explicitly targeting AI systems that generate or manipulate non-consensual intimate content.
In practical terms, tools creating synthetic nudes of identifiable people without their consent are now prohibited. The ban also covers the generation of child sexual abuse material.
The penalties provided are the highest in European AI law:
- Up to 35 million euros in fines for individuals or legal entities
- Or 7% of the annual global turnover for companies
- Legal enforcement is set to begin on December 2, 2026, with a transitional period for synthetic content labeling obligations
This framework changes the game for platforms that host or promote these services. Nudify sites and intimate deepfake apps will no longer be able to operate legally from Europe or target European users without facing severe legal repercussions.
Data Protection and Victim Recourse
Every photo submitted to a nudification tool constitutes personal data. The face is a biometric identifier. Submitting a photo of a third party without their consent violates GDPR, regardless of what the tool does with it afterward.
Nudification services also collect images sent by their users. What happens to these photos? Most of these platforms do not publish any credible privacy policy. The data may be stored, resold, or used to train other AI models.
For victims, several recourses exist. Reporting the content to the platform hosting it remains the first step. In France, the PHAROS platform allows for reporting illegal online content. Filing a complaint with the police or gendarmerie opens the way for criminal proceedings for invasion of privacy, distribution of intimate images without consent, or harassment.
Minors and their families can also contact the CNIL to assert the right to erasure of personal data used without consent.
Locking down photos on social media (private profile, restricting downloads) reduces the risk but does not eliminate it. Any photo published online can potentially be misused, making prevention as necessary as enforcement.
The gradual implementation of the European AI Act should force app stores and web hosts to remove these tools from their catalogs. The framework exists; its application will determine whether these services retreat or simply migrate to more permissive jurisdictions.