What the guidelines cover

• AI systems that interact directly with individuals, including chatbots and virtual assistants;
• AI-generated or manipulated audio, image, video and text;
• emotion-recognition and biometric-categorisation systems;
• deepfakes; and
• AI-generated text published to inform the public on matters of public interest.

Key points for businesses

Providers of interactive AI systems must generally design the system so that individuals are informed that they are interacting with AI, unless this is obvious from the circumstances. Providers of generative AI systems must also enable the detection of AI-generated or manipulated output through machine-readable marking.

Deployers may have separate disclosure duties where individuals are exposed to deepfakes, emotion-recognition or biometric-categorisation systems, or AI-generated public-interest content that has not undergone human review or editorial control.

The precise obligation therefore depends on the company’s role in the AI value chain. Providers are primarily responsible for technical design and marking measures, while deployers may be responsible for disclosures to users or the public.

Practical implications

The guidelines contain practical examples, definitions and exemptions, including for standard editing and situations in which the AI nature of an interaction is already obvious.

Companies that rely on the voluntary Code of Practice on Transparency of AI-Generated Content may use adherence to the Code as one means of demonstrating compliance. Companies that do not follow the Code must be able to demonstrate equivalent compliance through other appropriate measures.

What businesses should do now

1. Map covered systems. Identify chatbots, generative AI tools, emotion-recognition systems and other uses falling within Article 50.
2. Review user notices. Check when and how users are informed that they are interacting with AI.
3. Assess technical marking. Verify whether AI-generated or manipulated outputs are marked in a machine-readable and detectable format.
4. Create disclosure procedures. Establish rules for deepfakes and AI-generated public-interest content.
5. Update contracts. Allocate transparency, marking, evidence and cooperation obligations between AI providers, customers and deployers.

FAQ

When do the transparency obligations apply?

The Article 50 transparency obligations apply from 2 August 2026. Certain AI systems placed on the market before that date benefit from a limited transitional period for specific content-marking obligations.

Must all AI-generated content be visibly labelled?

No. The applicable requirement depends on the type of system, content and use. Providers may have machine-readable marking obligations, while deployers may have visible disclosure obligations in specified situations.

Are chatbots covered?

Yes. Individuals must generally be informed that they are interacting with an AI system unless this is obvious from the circumstances and context of use.

Sources

• European Commission, Guidelines on transparency obligations for providers and deployers of AI systems, 20 July 2026.
• Regulation (EU) 2024/1689, Article 50.