European Commission publishes guidance on AI Act transparency requirements

The European Commission has published guidelines to help organisations comply with transparency requirements under the EU AI Act. The requirements will begin to apply on 2 August 2026.

Providers will be required to design certain systems so that users are informed when they are interacting directly with an AI system. They must also add machine-readable markings to content generated or altered by AI so that it can be identified.

Organisations using these systems will have to inform people when they are exposed to deepfakes, emotion-recognition systems or systems that categorise people using biometric data.

Disclosure will also be required for AI-generated content concerning matters of public interest when it has not been subject to human review or editorial control.

The guidelines explain which systems and types of content fall within the rules. They cover chatbots, fully or partly generated text, deepfakes and other synthetic content. They also describe exceptions, including routine spelling and grammar corrections.

Organisations may demonstrate compliance by following a voluntary code of practice on the transparency of AI-generated content. The Commission and the European AI Board have recognised the code as an appropriate way to show compliance with the Act.

Most provisions of the AI Act, including enforcement powers for the Commission and national authorities, will apply from 2 August 2026. Systems already on the market before that date must meet the marking and detection requirements from 2 December 2026.