The European Parliament has approved a regulation prohibiting AI systems designed to produce nonconsensual intimate images, known as “nudifiers,” as part of a broader AI Act amendment. However, enforcement remains uncertain due to the lack of national authorities empowered to oversee compliance across most EU member states.
What Happened
On June 20, 2024, the European Parliament voted in favor of rules, following a provisional agreement from May 7 on the so-called “AI omnibus” legislation. This agreement includes a ban on AI technologies intended to generate explicit images of real identifiable persons without their consent. The regulation also requires companies to deploy safeguards against such content and could lead to market removal of offending systems by the end of 2024.
Key Facts
- The legislation targets AI systems that create nonconsensual sexual deepfakes involving real individuals, primarily women and children.
- The ban applies only to images involving identifiable persons under criteria set by the GDPR, such as face, name, or other distinguishing traits.
- Consent must be freely given, specific, informed, unambiguous, and explicit for intimate AI content to be legal.
- Providers must implement technical safeguards including prompt-level filters, data cleaning, abuse detection, and notice-and-action mechanisms.
- Fines for violations can reach up to €35 million or 7% of global annual turnover.
- Only eight EU member states currently have designated market surveillance authorities with powers to enforce the AI Act, with Italy and Denmark among the few enforcing nudity prohibitions.
- The full ban is scheduled to start on December 2, 2024, with broader high-risk AI rules delayed until August 2026.
Why It Matters
The regulation explicitly addresses a widespread digital safety abuse—nonconsensual AI-generated intimate images—that predominantly targets women and children. By holding both content creators and AI developers responsible, the EU aims to curb the industrial-scale abuse of AI for sexual image manipulation. However, enforcement delays risk leaving victims vulnerable despite the legislative ban.
Background
This policy builds on the EU’s Artificial Intelligence Act framework, first proposed in 2021, which categorizes AI risks and establishes regulatory requirements accordingly. The ban on “nudifiers” emerged after increasing public concern over sexual deepfakes and digital privacy violations, highlighted by cases such as nonconsensual AI-generated images of Italian Prime Minister Giorgia Meloni.
Analysis
Sadia Berdaï, head of Luxembourg’s Data Protection Authority AI Innovation Unit, emphasized that the ban focuses on “intended purpose” and use rather than banning technologies themselves. MEP Michael McNamara described the measure as necessary to industrialized abuse for profit. Digital rights advocate Belén Luna Sanz noted that while consent requirements strengthen protections, they also raise enforcement challenges by imposing higher thresholds on victims. MEP Sergey Lagodinsky indicated that most AI providers already have relevant safeguards, minimizing compliance costs.
Who Is Affected
- EU citizens, especially women and children, who risk being depicted in nonconsensual AI-generated intimate images.
- AI system developers and providers offering products capable of generating or manipulating intimate content.
- National regulatory authorities responsible for AI market surveillance within EU member states.
What Remains Unclear
- Specific enforcement mechanisms and responsibilities between national market surveillance authorities and the European AI Office.
- How differing cultural interpretations of “intimate content” will affect scope and enforcement.
- Exact procedures and timelines for national authorities to assume enforcement powers, as only a few member states currently have legal frameworks in place.
What Comes Next
Following the Parliament’s approval, the agreement awaits formal adoption by the Council and publication in the EU Official Journal. The ban on AI nudifiers is scheduled to come into effect on December 2, 2024. Broader high-risk AI provisions under the AI Act are delayed until August 2026.
Sources
This article is based on reporting and publicly available information from the following source:
Read more Digital Policy stories on Goka World News.