The EU is now taking concrete action against artificial intelligence abuse. The EU AI Act bans so-called nudifier apps – software that creates nude images of people without consent – and forces AI systems not to include such features in the first place. This is a direct strike against a growing problem: women in particular suffer from sexualized deepfakes that damage their reputation and dignity.
Key Facts
- Nudifier apps are banned: The EU AI Act prohibits the development and distribution of software that generates non-consensual sexualized images
- Labeling requirement for AI content: All AI-generated content must be clearly marked – intended to create transparency
- Sanctions must be "effective, proportionate and dissuasive": Violations can be punished with fines, warnings, or other measures (Art. 99 of the EU AI Act)
- Complaint channels for victims: Affected persons can file complaints with national supervisory authorities
Prevention Instead of Reaction
Sergey Lagodinsky, Green faction vice-chair in the European Parliament, explains the principle of the new regulation to the Frankfurter Rundschau: The EU AI Act tackles the problem "at its root". Rather than only compensating victims, tech giants must ensure during system development that abuse options don't arise in the first place.
"Nudifier apps have been banned. And AI that was not created for this purpose must ensure it does not include these options. This provides preventive protection."
Svenja Hahn (FDP) emphasizes in the same interview that the law respects the will of those affected: it only applies when someone becomes a victim of deepfakes against their will. Voluntary adult content remains permitted – it's a matter of self-determination.
Labeling Requirement: Transparency with Hurdles
The labeling requirement is meant to show users where AI is at work. This sounds reasonable – but leads to unexpected practical problems. According to Lagodinsky, subway operators have complained: if they switch their announcements to AI voices, they would need to attach a transparency notice to every single announcement. This is technically demanding and practically difficult to implement.
The regulation reveals a fundamental tension: how much transparency is necessary without stifling innovation?
Sanctions Insufficient – SPD Politician Demands More
Birgit Sippel (SPD) acknowledges the EU AI Act as an "important and correct step". However, she criticizes that existing sanctions – primarily fines – are insufficient to truly restrain tech giants. Her demand: time-limited market exclusions and strengthening European platforms as an alternative.
| Sanction Type | Description |
|---|---|
| Fines | Financial penalties |
| Warnings | Formal warnings |
| Non-monetary measures | e.g., market exclusion, operating bans |
What This Means for You
German companies developing or deploying AI systems must prepare for new compliance requirements. The labeling requirement affects not only tech giants but also mid-market firms integrating AI into their products. At the same time, the ban on abuse functions provides a clear guideline: whoever builds AI systems must eliminate abuse potential during the design phase. This costs time and resources – but is the price for a regulated European market where trust matters.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
