Brussels, Belgium – The European Union has begun enforcing new transparency rules requiring chatbots to identify themselves and deepfakes and other AI-generated content to be clearly labelled.

The measures, which took effect on Sunday, form part of the EU AI Act – described by European authorities as the world’s first comprehensive law governing artificial intelligence.

Under Article 50 of the law, people must be told when they are interacting directly with an AI system rather than a human.

Companies providing systems that generate or alter text, images, video and audio must also add machine-readable marks that help users and detection tools identify synthetic content.

Businesses and other organisations using AI must disclose deepfakes, emotion-recognition systems and biometric categorisation tools. AI-generated text for the public must also be labelled when it has not undergone human review or editorial control.

The rules apply to both AI developers and professional users. They could cover technology companies, banks operating customer-service chatbots, recruitment agencies using automated tools and media organisations publishing AI-generated material.

National market-surveillance authorities will lead enforcement, while the European AI Office will oversee certain systems linked to general-purpose AI models and large online platforms.

Violations can attract fines of up to €15mn or 3% of a company’s global annual turnover, whichever is higher.

A limited transition period applies to generative AI systems placed on the market before August 2. Material already published before the rules took effect does not require retrospective labelling.

The EU said the rules are intended to reduce deception, misinformation and manipulation as AI-generated material becomes increasingly difficult to distinguish from content created by people.