New EU Regulations on AI Content Labeling: What You Need to Know

Since August 2, a significant shift in how AI-generated content is handled has taken place within the European Union. Under new regulations, images, videos, and texts created with AI tools must now be clearly labeled to protect users from potentially misleading content. This initiative is part of the AI Act, the first comprehensive legislation globally aimed at regulating artificial intelligence.

The requirement for labeling applies to content used for professional purposes, such as online advertising and printed materials. Influencers who monetize their content are also obligated to comply with these new rules. Major AI providers, including OpenAI, Google, and Meta, are responsible for implementing the necessary labeling features.

The primary focus of these regulations is on deepfakes, which pose significant risks in areas like politics and consumer protection. For instance, online retailers could use convincingly realistic AI-generated videos of real individuals without customers being aware of the manipulation. According to EU guidelines, videos must be labeled at the beginning, with the label either permanently displayed or appearing at regular intervals.

For manipulated images, the EU suggests using interactive elements, such as a prompt that appears when users hover their mouse over a distorted face. Audio content, including AI-generated podcasts and phone chatbots, should feature a spoken disclaimer at the start. Texts are advised to include a notice indicating AI involvement, positioned either above the text or near the title.

Caution for Personal Use

While these regulations do not apply to private use, the line between personal and professional use is increasingly blurred. As Michael Terhaag, an IT law specialist, notes, individuals who regularly sell items on platforms like eBay or Vinted and use AI-generated images for product photos fall under these rules. Additionally, organizations that create posters with AI content must also adhere to the labeling requirements.

Violations of these regulations could lead to severe penalties. For instance, engaging in prohibited AI practices could incur fines of at least 35 million euros or seven percent of a company’s global annual revenue. Fines for failing to meet transparency requirements or lacking safety measures start at 15 million euros or three percent of annual revenue. Existing AI systems have a transition period until December 2, 2026, while new models must comply with the EU regulations immediately.