home_warning_banner Check now
Guide

Chatbots and AI: Article 50 transparency obligations

The AI Act imposes transparency obligations for chatbots, deepfakes and AI-generated content. Find out how to become compliant.

Author Thomas Bernard
January 5, 2025 10 min read
Chatbots and AI: Article 50 transparency obligations
Chatbots must clearly identify themselves as AI systems

Even if your chatbot is not classified as "high-risk", Article 50 of the AI Act imposes specific transparency obligations. The goal: to ensure that users always know when they are interacting with AI.

Why is transparency crucial?

The European legislator considered that certain uses of AI, even non-risky ones, require that people be informed. This helps preserve their decision-making autonomy and prevent manipulation.

Natural persons should be notified that they are interacting with an AI system, unless this is obvious from the circumstances.

Recital 132 of the AI Act

This obligation applies from August 2026 and covers a wide range of systems.

Which systems are affected?

Article 50 targets three types of systems:

Chatbots and assistants

Any system intended to interact directly with natural persons.

Image generation

Systems that generate synthetic images, audio or video.

Deepfakes

Content that manipulates or generates content resembling real people.

Detailed obligations

Obligations vary by system type:

For chatbots

Deployers must ensure that people are informed they are interacting with an AI system, unless it is obvious given the circumstances and context of use.

Example of compliant message
I am a virtual assistant powered by artificial intelligence. How can I help you?

The message must be clear, visible and understandable to the average user.

For deepfakes and generated content

Obligations are stricter for synthetic content:

  • Mention that the content has been artificially generated or manipulated
  • Use machine-readable marking when technically feasible
  • Exception for artistic, satirical or fictional content clearly presented as such

How to become compliant?

Here is a summary of actions to implement based on your system:

System type Obligation How to do it?
Customer chatbot AI interaction notice Welcome message + mention in interface
Image generator Content marking Visible watermark + metadata
Voice synthesizer Synthesis notice Oral mention + file marking

Key points to watch

Some aspects deserve increased vigilance during compliance:

Practical advice

Integrate notification at the design stage (privacy by design). It's easier than modifying an existing system and demonstrates your commitment to compliance.

Article 50 compliance checklist

Check your transparency compliance

Our audit evaluates whether your systems meet Article 50 obligations.

Start free audit

Transparency is not just a legal obligation: it's also a guarantee of trust for your users. By clearly communicating about AI use, you strengthen your credibility and prepare for the public's growing expectations regarding digital ethics.

Share:

Related articles

Stay informed

Receive our AI Act analysis and guides directly in your inbox.