Greater transparency from AI providers and deployers: the EU Commission’s new guidelines

Brussels – “With today’s guidelines, the Commission is supporting the smooth and effective implementation of the European Artificial Intelligence Regulation, to make AI content and systems that interact with people, such as chatbots and AI agents, more transparent and trustworthy.” This was stated today (20 July) by the European Commission’s Executive Vice-President for Technological Sovereignty, Henna Virkkunen, on the occasion of the publication of the executive’s guidelines to help providers and developers of certain artificial intelligence systems comply with the transparency obligations set out in the Artificial Intelligence Act, which will come into force on 2 August. Virkunnen emphasised that, in addition to encouraging “providers and developers to comply with their obligations under the European AI Regulation,” the guidelines will help citizens “understand when they are interacting with AI.”

According to the Commission, “the rapid development of generative and interactive AI systems is making it increasingly difficult to distinguish between AI-generated interactions and content and authentic, human-created content.” Not only that, but “individuals are increasingly exposed to emotion recognition and biometric categorisation without their knowledge.” This entails “new risks of large-scale disinformation and manipulation, fraud, identity theft, and consumer deception.” 

In this context, the guidelines for suppliers, AI system developers and the relevant authorities aim to ensure compliance with Article 50 of the European AI Regulation, which requires that users are always informed when interacting with an AI, and that artificially generated content is clearly recognisable.

The guidelines therefore stipulate that providers of AI systems must ensure that “individuals are explicitly informed whenever they interact directly with an AI system” and must include “machine-readable markers to enable the detection of AI-generated or manipulated content.” Meanwhile, those implementing these systems are required to inform users when they are exposed to emotion recognition and biometric categorisation tools; deepfakes (deceptively realistic multimedia content often generated using AI); and texts on topics of public interest that are published without human review or editorial control.

In a few days, on 2 August, most of the provisions set out in the European Regulation on Artificial Intelligence will come into force, including the powers of the European Commission and national market supervisory authorities to impose sanctions. AI systems placed on the market before August will have to comply with the labelling and tracking requirements from 2 December 2026. 

To help organisations better understand their obligations under the European AI Regulation, in addition to the guidelines, the Commission is also providing other tools such as the Code of Practice on Transparency of AI-Generated Content, a document drawn up by independent experts with input from hundreds of stakeholders, and the AI Act support service, a digital helpdesk launched by the European Commission and the European AI Office to support businesses, start-ups, public bodies and citizens in applying the Regulation. Although adherence to the code is voluntary, the transparency requirements set out in Article 50 are legal obligations. The Commission and the European AI Committee have confirmed that the code is an appropriate voluntary instrument to demonstrate compliance with the Regulation’s transparency obligations. Finally, among other tools, the European Union has created a set of icons that creators, publishers, and other operators of generative AI systems can use to label their AI-generated content.