Key Takeaways
- Starting 2 August 2026, the EU will enforce mandatory transparency rules for certain AI systems, requiring clear labelling of AI‑generated or manipulated content and explicit disclosure when users interact with an AI rather than a human.
- The rules apply to deep‑fake‑style media (images, audio, video that resemble real persons, places or events), emotion‑recognition and biometric‑categorisation tools, and publicly‑distributed text that lacks human editorial oversight.
- Providers and deployers must use the EU‑designed set of icons and embed machine‑readable marks so that the labelling is both visible and technically detectable.
- When an AI system—such as a chatbot, AI agent, or avatar—is used, users must be plainly informed that they are not conversing with a real person.
- The European Commission has issued detailed guidelines and a code of practice to help organisations demonstrate compliance, including technical standards for marking and procedural steps for user notification.
- Enforcement falls to national market‑surveillance authorities, the European AI Office (for systems under its remit), and the European Data Protection Supervisor (when EU institutions are involved); violations can attract fines of up to €15 million or 3 % of global annual turnover for companies, and up to €750 000 for EU bodies, with proportionality considerations for SMEs and SMCs.
- These transparency obligations are part of the broader AI Act, which entered into force on 1 August 2024 and is being rolled out in stages; the Act aims to create a harmonised, trustworthy‑AI single market while protecting health, safety, fundamental rights, democracy and the rule of law.
- Effective implementation is now a top priority for the Commission, which encourages stakeholders to consult the press releases, FAQs and quick‑facts documents for practical guidance.
The Growing Need for AI Transparency
Artificial intelligence is advancing at a pace that makes it increasingly difficult to distinguish AI‑generated and manipulated content from human‑created, authentic material. This blurring raises significant societal risks, including large‑scale misinformation, fraud, impersonation and consumer deception. Recognising these dangers, the EU has introduced a set of transparency obligations designed to empower individuals to recognise when they are dealing with AI or AI‑produced content, thereby enabling informed decisions and stronger protection against deception.
Effective Date and Scope of the New Rules
The transparency obligations officially take effect on 2 August 2026. They apply to providers and deployers of specific AI systems that produce or disseminate content capable of misleading audiences. The scope is deliberately focused on high‑risk categories where the potential for harm is greatest, ensuring that the regulatory burden remains proportionate while addressing the most pressing threats to public trust.
Marking and Labelling of AI‑Generated Content
One cornerstone of the new regime is the requirement that certain AI‑generated or manipulated media be clearly and visibly labelled and include machine‑readable marks. As the original text states, “The EU has created a set of icons that can be used for this purpose.” This labelling mandate covers:
- Images, audio and video that resemble existing persons, objects, places, entities or events (commonly known as deepfakes).
- Emotion‑recognition and biometric‑categorisation tools, which analyse or infer personal characteristics from biometric data.
- Text published to inform the public on matters of public interest when there has been no human review or editorial control.
By obliging both a visible icon (for immediate human perception) and a machine‑readable tag (for automated detection), the rule aims to make AI‑origin unmistakable across different consumption channels—whether a viewer scrolls social media, watches a news broadcast, or interacts with an online advertisement.
Transparency When Interacting with an AI System
Beyond content labelling, the rules also demand that users be clearly informed when they are not interacting with a real person, but with an AI system such as a chatbot, AI agent, or avatar. This provision seeks to prevent scenarios where individuals unwittingly disclose sensitive information or form opinions based on advice they believe stems from a human expert. The requirement is straightforward: any interface that simulates human conversation must disclose its artificial nature at the outset of the interaction, using language that is unambiguous and readily noticeable.
Guidelines, Code of Practice and Demonstrating Compliance
To assist organisations in meeting these obligations, the European Commission has published comprehensive guidelines that explain how compliance can be demonstrated. The guidelines recommend adherence to a code of practice that outlines technical standards for the icons and machine‑readable marks, as well as procedural steps for user notification. By following the code, providers can generate tangible evidence—such as labelling logs, audit trails, and user‑interface screenshots—that they are in line with the legal expectations, facilitating smoother oversight by enforcement bodies.
Enforcement Authorities and Potential Sanctions
Responsibility for enforcing the transparency rules is shared among several entities: national market‑surveillance authorities, the European AI Office (for systems under its supervision), and the European Data Protection Supervisor (when EU institutions act as providers or deployers). These authorities are empowered to investigate suspected breaches and impose penalties. According to the source, fines may reach up to €15 million, or 3 % of global annual turnover for companies, while EU bodies face a maximum of €750 000. The framework also incorporates proportionality, meaning that small and medium‑sized enterprises (SMEs) and small mid‑cap companies (SMCs) will see fines calibrated to their size and economic capacity, ensuring that the regime is both deterrent and fair.
Relation to the AI Act and Implementation Timeline
The transparency obligations form a critical component of the wider AI Act, which entered into force on 1 August 2024. The Act is being rolled out in stages, with different sets of requirements becoming applicable at prescribed intervals. By establishing a single market and harmonised rules for trustworthy AI, the legislation seeks to foster innovation while safeguarding fundamental rights, health, safety, democracy and the rule of law. The Commission now regards the effective implementation of these transparency rules as a key priority, urging stakeholders to monitor upcoming guidance and prepare their systems accordingly.
Outlook and Recommendations for Stakeholders
As the 2 August 2026 deadline approaches, businesses developing or deploying AI‑generated media, emotion‑recognition tools, or public‑interest text should begin integrating the EU’s icon set and machine‑readable marking protocols into their production pipelines. Likewise, any service that employs conversational AI must update its user interfaces to include clear, upfront disclosures about the non‑human nature of the interaction. Consulting the Commission’s guidelines and the associated code of practice will not only reduce the risk of costly sanctions but also reinforce consumer trust—a competitive advantage in an market increasingly wary of AI‑driven deception. By embracing transparency now, organisations can align with the EU’s vision of an AI ecosystem that is both innovative and accountable.
https://commission.europa.eu/news-and-media/news/safer-and-more-transparent-ai-2026-08-02_en