The implementation of the European Artificial Intelligence Act (AI Act) is entering a critical new phase. Starting August 2, specific provisions aimed at enhancing transparency and accountability will take effect, directly impacting tech providers and any business integrating AI tools into its operations.
Transparency and Content Labeling
The new framework mandates that AI system providers clearly inform users when they are interacting with machines, such as chatbots and digital assistants. A significant focus is placed on deepfakes and AI-generated or altered content (text, images, audio, video), which must now carry specific transparency labels.
The goal is to curb misinformation and public manipulation. However, the European Commission clarifies that simple assistance functions, such as automated spell-check or grammar tools, are exempt from the strict labeling requirements applied to fully synthetic content.
The Greek Implementation Framework
Greece has already established its national framework, identifying the competent authorities and oversight mechanisms:
- Hellenic Data Protection Authority (HDPA): Designated as the central market supervisory authority.
- EETT (Hellenic Telecommunications and Post Commission): Acts as the notifying authority and host of the Coordination and Expertise Center for AI.
- AI Registry: A unified registry will be created for AI systems utilized by Public Administration.
- AI Regulatory Sandboxes: Controlled environments will be established for testing innovative applications under real-world conditions.
A notable feature of the Greek legal framework is the introduction of criminal penalties for the intentional removal of mandatory transparency labels from deepfake content, alongside the administrative sanctions provided for other violations.
