EU AI Transparency Rules Take Effect in 2026: What Businesses Must Disclose

The EU AI Act, set to take effect from August 2, 2026, introduces new transparency requirements for businesses that provide or deploy certain AI systems. Article 50 of the regulation mandates disclosure when individuals interact with AI and establishes labeling standards for AI-generated or manipulated content, including deepfakes. The obligations apply to companies utilizing chatbots, generative media tools, emotion-recognition systems, and biometric categorisation.

Businesses must identify where AI is visible to individuals and make its role clear. The rules aim to prevent deception and enable people to recognize when content or interactions involve AI. They also provide a concrete implementation deadline for tool providers and organizations incorporating AI into customer, marketing, editorial, or public-facing workflows.

Article 50 divides transparency duties between providers, who develop or place AI systems on the market, and deployers, who use these systems. The specific duty depends on the type of system and its usage. When a person interacts with an AI system, providers must inform individuals that they are interacting with AI. If an AI system generates or manipulates content, providers must make the artificial origin of the content verifiable through machine-readable marking.

Deployers have their own disclosure responsibilities when they publish deepfakes or AI-generated public-interest text. The EU has also released a voluntary Code of Practice on Marking and Labelling AI-Generated Content to provide a practical framework for compliance. This code encourages open standards for content marking and includes an EU icon set for consistent labeling.

Businesses should assess actual user journeys and outputs rather than relying on vendors' general descriptions. Content labels and deepfake disclosures will be the most visible changes. For example, a business may need to disclose when a chatbot or generative media tool is being used, or when AI-generated content is published, especially deepfakes or AI-produced text with public interest.

The deadline for these obligations is August 2, 2026, although older generative AI content may extend the deadline to December 2026. Businesses should create an inventory of AI systems, identify potential user interactions, review generated content, and work with vendors to understand their support for machine-readable content marking.

By documenting their processes and configurations, businesses can reduce the need for last-minute changes and improve customer clarity. Non-compliance could result in penalties of up to €15 million or 3% of global turnover, underscoring the importance of proactive compliance.