According to the portal BornCity, French regulators have started first compliance checks. They are aimed, among others, at professional users who must mark synthetic media and deepfakes with visible labels and machine-readable markers. The checks take place as part of an expansion of national oversight.
Transparency obligations under Article 50
According to BornCity, the focus lies on applying central transparency rules: clear labelling of chatbots, watermarking of generative media and disclosure obligations for deepfakes. The source states that the transparency obligations under Article 50 of the AI Act have applied since 2 August 2026. Providers face significant penalties for serious breaches. Accompanying guidance from regulators says users must be clearly informed when they interact with automated systems or consume synthetic content on matters of public interest.
Limitation period at EU level
At European level, the European Commission presented on 12 October 2026 a draft implementing regulation (Ares(2026)9589403). Under it, the European AI Office would receive a limitation period of at most ten years to pursue serious breaches by general-purpose AI models such as ChatGPT and Gemini. The authority would also gain extensive access to internal documents, while the protection of trade secrets must be preserved. A public consultation on the draft runs until 9 November 2026. According to the source, major model providers such as OpenAI and Google DeepMind did not comment on the planned deadlines.
High-risk timeline and standards
According to the source, the timeline for high-risk systems remains in motion. The further obligations were postponed via the digital omnibus: to 2 December 2027 for systems under Annex III and to 2 August 2028 for systems embedded in regulated products under Annex I. According to BornCity, the first harmonised standard, EN 18286, was published in the EU Official Journal on 14 October 2026; however, it only covers the presumption of conformity for Article 17 on quality management systems, not cybersecurity, risk management or human oversight.
Assessment
For companies using synthetic media or chatbots in Europe, the labelling obligation under Article 50 could become relevant in the short term if national authorities expand their checks. According to the source, it is also not yet decided whether the planned limitation rule and the consultation on incident reporting, running until 7 November 2026, will change the legal position.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
