The Core Transparency Obligations (Article 50)
Article 50 of the EU AI Act mandates that specific AI systems must be transparent in their operations. The goal is to prevent deception and ensure that natural persons are aware they are dealing with artificial intelligence.
This applies regardless of whether the system is classified as high-risk. If a system mimics human interaction, generates content, or categorizes individuals, transparency measures must be visibly integrated into the user experience.
Which AI Systems Require Transparency?
Transparency notices are mandatory for AI systems designed to interact directly with humans, such as customer service chatbots or virtual assistants. Users must be informed promptly and clearly that they are communicating with a machine.
Obligations also apply to emotion recognition systems and biometric categorization systems, where users must be informed of the operation of the system before processing begins.
- Chatbots and conversational agents
- Emotion recognition software
- Biometric categorization tools
- Deepfake and synthetic content generators
Deepfakes and AI-Generated Content Disclosures
A major focus of the transparency rules is synthetic media. Deployers who generate or manipulate image, audio, or video content that significantly resembles existing persons, places, or events (deepfakes) must disclose that the content is artificially generated.
Similarly, AI systems that generate general text, such as news articles or informational summaries published for public consumption, must be labeled as machine-generated, unless subject to human editorial review and responsibility.
Why You Need an Internal AI-Use Register
To manage these obligations, organizations should maintain an internal AI-Use Register. This register tracks all AI systems deployed across the company, their risk classification under the EU AI Act, and the transparency measures applied.
A comprehensive register provides an oversight record for auditors, proving that the organization actively monitors its AI deployments and complies with Article 50 disclosure requirements.
Frequently asked questions
Do we need to declare AI usage if it is obvious to the user?
Exceptions exist if it is undeniably obvious from the circumstances and context of use that the user is interacting with an AI system, though providing a notice is always the safer compliance route.
Who is responsible for the transparency notice: the developer or the user?
The obligation generally falls on the 'deployer' (the company using the AI to interact with customers) to ensure the notice is provided, though 'providers' (developers) must design the system to enable this transparency.
Does AI-generated art need a label?
If the art is generated by a generative AI model, standard practice under the Act requires it to be marked in a machine-readable format indicating its synthetic origin, and clear disclosure is required for deepfakes.
What constitutes an AI-Use Register?
It is an internal log documenting the name, purpose, risk tier, data inputs, and oversight mechanisms for every AI system used within an organization to ensure governance and compliance.