Art. 50 EU AI Act — Transparency: inform users about AI interaction

Among all EU AI Act obligations, Art. 50 is the next to take effect for most organisations: 2 August 2026. For organisations deploying chatbots, recommendation systems, emotion recognition or synthetic content, this is not a theoretical obligation — it is a deadline two months away.

This article explains what Art. 50 requires, which systems it covers, what exceptions exist and what the concrete steps are to comply on time.

What is Art. 50 and who does it apply to?

Art. 50 contains transparency obligations for AI systems that interact directly with people or generate content that could be mistaken for real. The law distinguishes three groups of obligated parties:

• Providers: organisations that place AI systems on the market — they must technically enable systems to fulfil the information obligation
• Deployers: organisations using AI systems in their processes — they are responsible for actually informing users
• Persons distributing synthetic content: anyone publishing AI-generated images, audio or video to the public

For most organisations, the deployer role is most relevant.

Which systems fall under Art. 50?

Art. 50 targets four categories of AI applications:

1. Chatbots and conversational AI (Art. 50.1)
Any AI that communicates with people and could be mistaken for a human must inform the user that they are interacting with an AI system — unless this is evident from context. This applies to customer service bots, HR assistants, digital coaches, legal information systems and similar applications.

2. Emotion recognition systems (Art. 50.3)
AI that detects emotions or psychological states of individuals must inform those persons. Think of video conferencing software with emotion analysis, HR tools interpreting candidate behaviour or customer service AI applying sentiment analysis.

3. Biometric categorisation (Art. 50.3)
AI classifying persons based on biometric characteristics must inform those persons — unless used for lawful identity verification purposes.

4. Synthetic content (Art. 50.4)
AI-generated or AI-manipulated images, audio and video (deepfakes) must be labelled as AI-generated. Exception: content for artistic, creative or satirical purposes with a clear artistic context.

What must you communicate — and how?

The information obligation has two dimensions: what you communicate and when you communicate it.

What: Users must know they are interacting with an AI system. You do not need to fully disclose how the system works — but the AI nature may not be hidden or actively denied.

When: The information must be provided at the moment of or before the interaction. Informing after the fact is insufficient. A footnote in the privacy policy is insufficient. The communication must be active, comprehensible and timely.

How: The law prescribes no specific format. Common approaches include: a visible badge ("You are speaking with an AI assistant"), an opening message when starting a chat interaction, or a clear visual indication on AI-generated content. The key is that the average user understands it without legal training.

Exception: internal business processes

Art. 50.1 includes an important exception: the information obligation does not apply when it is evident to the natural person that they are interacting with an AI system. This is relevant for internal applications where employees explicitly work with an AI tool and are aware of its AI nature.

But note: "evident" carries a high burden of proof. If there is any doubt about whether the employee knows and understands this, the information obligation still applies. Document explicitly why you consider an application to fall under the exception.

Relationship to Art. 26.7 — Deployer transparency towards those affected

Art. 50 is not the only transparency obligation. Art. 26.7 requires deployers using high-risk AI to inform affected persons about the deployment of that system. This goes further than Art. 50: it also covers systems where no direct interaction occurs but where a person is directly assessed (such as CV screening or credit scoring).

For high-risk AI, you must simultaneously comply with both: Art. 50 (interaction transparency) and Art. 26.7 (assessment transparency). They complement each other.

Deadline and enforcement

Art. 50 enters into force on 2 August 2026 — alongside the broader transparency provisions of the EU AI Act. After that date, supervisory authorities can initiate enforcement. The maximum fine for violating Art. 50 is €15 million or 3% of global annual turnover.

Beyond direct enforcement, there is the risk of complaints from consumers and affected individuals via national supervisory authorities.

Practical steps for compliance before 2 August 2026

Step 1 — Inventory all AI systems with direct user interaction. List all chatbots, recommendation engines, emotion analysis systems and synthetic content tools your organisation uses or publishes. Don't forget built-in AI in existing software — think of AI functions in CRM, customer service platforms or communication tools.

Step 2 — Assess per system whether the information obligation applies. Does the exception apply (evident to user)? Or is informing required? Document the reasoning.

Step 3 — Implement the information disclosure. Ensure the AI nature is visible at the moment of interaction. Test whether the average user understands it. Document what you have implemented and when.

Step 4 — Coordinate with your vendors. Providers are required to make their systems technically suitable for Art. 50 compliance. Verify with your SaaS vendors whether they have implemented the required functionality. Contractually establish who is responsible for which part of the information provision.

Step 5 — Document for your compliance dossier. The supervisory authority may request evidence of compliance. Save screenshots, process descriptions and vendor confirmations. A compliance dossier with timestamps is your best defence.

Two months

Art. 50 is not a heavy technical implementation — but it does require action. Most organisations need to review and update their chatbots, customer service systems and content pipelines. Two months is tight if vendors also need to act. Start now.