From Sunday, companies operating in the European Union must clearly label any content generated or manipulated by artificial intelligence, including deepfakes and chatbots. The new rules, part of the bloc's sweeping AI Act, aim to help citizens distinguish between authentic and synthetic material online.
The requirements apply to AI systems such as chatbots, which must now inform users at the start of a conversation or through a persistent on-screen label that they are interacting with a machine. The disclosure must be obvious to a typical user, not buried in terms of service. Any AI-generated image, video, or text must carry a visible label, either via embedded watermarks or other machine-readable markers that enable easy detection.
“Generative AI enables disinformation to be created at an unprecedented scale, tailored to specific audiences, and disseminated with remarkable speed,” an EU official said. The official added that AI is “making it increasingly difficult for all of us to distinguish what is real from what is synthetic,” and that the rules aim to “preserve citizens’ ability to trust what they see, hear and read.”
Scope and exemptions
The rules cover content produced for professional purposes. People using AI in a purely personal capacity are not affected. Text intended to inform the public on matters of general interest must also be labelled if it was created by AI without human editorial oversight. Exemptions apply to “artistic, creative, satirical [and] fictional” works.
Existing AI systems have until 2 December to adapt. The EU has faced criticism from some business groups for adding regulatory burdens, but Ashley Casovan of the International Association of Privacy Professionals noted that compliance challenges are often overcome: “We have heard that it is going to be very, very difficult to implement. But I think we often hear this with compliance requirements. And yet, the world turns and we figure these things out.”
The rules are part of a broader EU push to regulate AI, which also includes mandatory deepfake labels under the AI Act, though enforcement remains a challenge.
Big Tech prepares
Major technology firms have already begun applying their own labels ahead of the EU deadline. TikTok has required creators to label AI-generated images, audio, and video for several years, and says more than 3 billion pieces of content already carry labels using its detection tools. Meta has introduced an “AI Info” label on Instagram and Facebook for AI-generated posts. Google has signed the EU’s code of conduct on AI transparency and says it is working with Nvidia, OpenAI, and Apple on digital tagging tools.
However, Google’s Karen Massin warned of “regulatory complexity” that could prove counterproductive and “risks confusing the people these rules are meant to help.” She added: “If online content is flooded with overlapping AI labels and legal disclosures, it becomes harder for people to get the clear context they need.”
The rules also come amid concerns about AI’s role in spreading propaganda. A recent investigation found that AI chatbots unwittingly amplify Russian propaganda from EU-sanctioned outlets, highlighting the urgency of transparency measures.
Companies that fail to comply face significant fines. The EU says the rules are designed to protect democratic discourse and public trust in the digital information ecosystem. As the technology evolves, the bloc plans to update the requirements periodically.
