A new European Union law has taken effect requiring clear disclosures when content or customer service tools are generated by artificial intelligence rather than humans.
The regulation targets digital media where non-human involvement might not be immediately obvious, including realistic image generations, synthesized voice audio, deepfakes, and automated phone or chat assistants.
Watermarks and visible labels
Under the new rules, artificial intelligence outputs must feature explicit markings, such as a prominent banner reading “AI GENERATED” or digital watermarks embedded directly into photo and video files.
Companies operating automated customer service lines or AI receptionists are now expected to inform users at the beginning of interactions via spoken messages or text notifications.
The requirements specifically focus on media connected to the public interest, such as medical advice, official government communications, and synthetic depictions of real individuals.
Focus on transparency and trust
European officials designed the framework to build public trust in digital media as synthetic content becomes increasingly sophisticated and common across popular online platforms.
Research conducted earlier this year by the University of Amsterdam highlighted the scope of the issue, finding that roughly nine out of ten AI-generated images used during municipal elections lacked any clear identification label.
Informal uses, such as basic promotional flyers for local events or commercial merchandise displaying generated artwork, remain exempt from these mandatory labels for the time being.
Enforcement and potential penalties
Organizations that fail to comply with the new transparency rules face significant financial penalties, which can reach up to €15 million following official warnings.
Although the regulation is officially active, formal administrative enforcement in the Netherlands is currently pending as the Autoriteit Persoonsgegevens awaits an official mandate from the national government.
The regulatory authority previously raised issues regarding automated technology and digital safety, similar to when the Dutch privacy regulator warned period-tracking app users about data risks.
While full administrative oversight from the regulator is expected to begin next year, non-compliant entities can already face legal challenges through the court system.
