From Sunday, EU rules require companies to label AI-driven chatbots and any AI-generated text, images, audio or video. The rules allow watermarks or other markers and set a December 2 deadline for existing systems to comply. Exemptions cover personal use and clearly artistic, satirical or fictional works. Major platforms like TikTok, Meta and Google are already adding labels, though some warn that overlapping disclosures could confuse users.
EU Requires Clear Labels on AI-Generated Content From Sunday — Deepfakes Must Be Marked
From August 2, companies must make sure their AI systems like chatbots make it clear to users they are AI, or when an image or text has been created using AI (Nicolas TUCAT)
The European Union's new transparency rules for artificial intelligence begin to take effect from Sunday, requiring companies to clearly label AI-driven systems and identify text, images, audio or video produced or manipulated by AI.
What Changes From Sunday?
Under the phased rollout of the EU's comprehensive AI law, businesses must now ensure that interactive systems such as chatbots explicitly disclose they are AI, and that any image, text or audio generated by AI carries a visible label. Labels can include embedded watermarks or other digital markers designed to make synthetic content easy to detect. Firms that fail to comply face substantial fines.
Why The Rules Matter
“Generative AI enables disinformation to be created at unprecedented scale, tailored to specific audiences, and disseminated with remarkable speed,” an EU official said, adding that the rules are intended to preserve citizens’ ability to trust what they see, hear and read.
The EU is particularly concerned about deepfakes — text, images, video and audio that appear authentic but are generated or altered by AI. The rules aim to curb misuse that could mislead the public, manipulate elections, or cause reputational harm.
Scope, Deadlines And Exemptions
The rules apply primarily to content created for professional or public-interest purposes. Individuals using AI solely for private, personal reasons are generally exempt. Text intended to inform the public about matters of general interest also must be labeled when produced solely by AI without human editorial oversight. Existing AI systems are given until December 2 to adapt to these requirements, and exemptions have been carved out for clearly artistic, creative, satirical or fictional works.
Industry Response
Major technology companies have already begun adding labels in anticipation of the rules. TikTok says it has required creators to tag AI-generated images, audio and video for years and claims more than three billion items have labels. Meta displays an “AI Info” label across Facebook and Instagram, while Google has signed the EU code of conduct on AI transparency and is collaborating with companies such as Nvidia, OpenAI and Apple on digital tagging tools.
Google’s Karen Massin warned of potential "regulatory complexity" that could backfire by flooding users with overlapping labels and legal disclosures, making it harder for people to get clear context.
Other experts say implementation will be challenging but feasible. Ashley Casovan of the International Association of Privacy Professionals noted that compliance often seems difficult at first, but organizations typically adapt.
What To Watch Next
Enforcement and the development of interoperable tagging standards will be critical. Regulators, tech firms and civil society will need to coordinate on robust, user-friendly labeling methods to ensure the rules meet their transparency goals without creating confusing or redundant disclosures.
Help us improve.
