The European Union began enforcing its AI Act transparency obligations on August 2, requiring companies to disclose when users are interacting with artificial intelligence and to label AI-generated content, with penalties reaching up to €15 million or 3% of global turnover.
The rules, contained in Article 50 of the landmark AI Act, apply to providers and deployers of AI systems across four main areas: direct interaction disclosure, synthetic content labeling, emotion recognition transparency, and deepfake identification. Companies had a six-month transition period after the act entered force in February.
On July 20, the European Commission adopted dedicated guidelines to clarify the scope and application of the obligations. The guidelines specify that providers of systems generating synthetic audio, images, video, or text must implement technical solutions such as machine-readable markings so content can be detected as AI-generated.
For generative AI systems already on the European market before August 2, a transitional deadline of December 2, 2026, applies for the machine-readable marking requirement. Deployer obligations, however, took effect immediately on August 2.
National market surveillance authorities and the Commission's AI Office are responsible for oversight. The European Data Protection Supervisor will monitor compliance within EU institutions. A Code of Practice on AI-generated content is under development to support practical implementation.
Legal experts warn that the fines, while substantial, are only part of the risk. Reputational damage from non-compliance could be more costly, particularly for consumer-facing platforms where trust is paramount. Major tech companies have been quietly updating their systems to meet the requirements ahead of the deadline.
Image source: i.ibb.co