EU now requires companies to label AI content
New artificial intelligence transparency rules took effect on August 2, 2026, aiming to protect citizens from deepfakes and disinformation.
New artificial intelligence transparency rules took effect on August 2, 2026, aiming to protect citizens from deepfakes and disinformation.
As of Sunday, August 2, 2026, a new phase of the comprehensive European AI Act has come into force. The key change: companies must now clearly label AI-generated content, including deepfakes, so that users immediately know whether what they see, hear, or read is real or synthetic.
According to Euronews, the goal is to preserve citizens' trust in the digital space. "Generative AI enables the creation of disinformation on an unprecedented scale, tailored to specific audiences and spread at incredible speed," an EU official told Euronews. He added that AI "makes it increasingly difficult for all of us to distinguish what is real from what is synthetic."
The rules apply to professional use of AI and do not affect individuals using it for personal purposes. They cover chatbots that must disclose they are AI, labels on AI-generated images and texts, and informing individuals when they are exposed to emotion recognition and biometric categorization tools. Existing AI systems have until December 2, 2026, to adapt, with exemptions planned for artistic, creative, satirical, and fictional works.
Tagesschau reports that companies failing to comply with the new rules face serious fines. These can reach up to €15 million or three percent of their global annual turnover. However, enforcement will not be straightforward. Ashley Casovan of the International Association of Privacy Professionals (IAPP) told Euronews: "We've heard that this will be very, very difficult to enforce. But I think we often hear that with compliance requirements. Nevertheless, the world turns, and we manage."
On the other hand, Google's Karen Massin warns of "regulatory complexity" that could be counterproductive. "If online content is flooded with overlapping AI labels and legal disclaimers, it becomes harder for people to get the clear context they need," she told Euronews.
Tagesschau also brings reactions from citizens in Cologne, where passersby were asked about their ability to distinguish real images from AI material. "I simply feel underprepared!" one respondent said. Another noted: "AI images are usually too beautiful, too perfect-and that makes me suspicious." A third passerby added: "Sometimes I'm surprised at how quickly this has developed, that you almost always have to question whether it's AI."
Tech giants have already begun introducing their own labels in anticipation of the regulation. Euronews notes that TikTok, thanks to detection tools and technology, has already labeled three billion pieces of content, while Meta has activated the "AI Info" label on Instagram and Facebook. Google has signed the EU's AI transparency code of conduct and is collaborating with Nvidia, OpenAI, and Apple on digital labeling tools.
Professor of business law Daniel Graewe believes that concerns about overregulation have subsided given recent developments. "We have to say, honestly, the AI Act initially received quite strong negative feedback. It was said that the EU was regulating a new economic area to death before it even started growing. Now, seeing where all this with AI is heading, we can be grateful, especially given the increasing disinformation through AI," he told Tagesschau.
However, Graewe is skeptical about the global reach of European rules. In an interview with the ARD Europa-Podcast punktEU, he stated: "Not much will happen there, for sure. The eternal contradiction: in Europe, we have certain rules, but they don't apply globally on the internet. We won't resolve that contradiction with these transparency rules either."
The new regulations come at a time of heightened global debate over AI safety. Tagesschau recalls recent incidents, including a case where an AI agent escaped during a security test and independently hacked another company. Elon Musk commented that AI will surpass humans within five years, while OpenAI CEO Sam Altman stated: "Wir haben Singularität bereits erreicht" (We have already reached the singularity). The American company Anthropic also reported that its AI model, during tests, unexpectedly penetrated other companies' systems.
