On this page
Published by horak. Attorneys at law / Patent attorneys
2026 legal update / Reviewed 12 November 2026
The AI Act’s transparency rules apply from 2 August 2026. Product teams need to distinguish visible user information from technical marking duties and assess the exceptions separately.
Separate the obligations
The Commission’s July 2026 guidance addresses interactions with AI, certain biometric uses and synthetic content. Provider-side machine-readable marking and deployer-side disclosure are different duties. A banner on the website is not necessarily a substitute for the technical requirement.
Check the transition narrowly
The Commission describes a limited transition until 2 December 2026 for Article 50(2) marking and detection obligations concerning systems placed on the market before 2 August 2026. This is not a general grace period for all transparency obligations. Public-interest text, human review and editorial responsibility require their own analysis.
Four practical examples
Illustrative scenarios, not reports of client mandates. The appropriate response depends on the facts and applicable law.
Chatbot interface
A business checks whether users receive the required information at the relevant interaction, including mobile and embedded versions.
Synthetic executive video
A company assesses whether a realistic generated video is a deepfake requiring disclosure and obtains any necessary personality-rights permissions.
Legacy generation service
A provider documents when its system entered the market before relying on the narrow marking transition.
Editorial workflow
A publisher records human review and editorial responsibility, while assessing separately any obligations affecting images or other synthetic material.
Practical next step
Identify the systems, products or contracts affected, assign a responsible owner and preserve the documents supporting your assessment. Read our related service page for the scope of legal support.