EU AI Act Transparency Obligations: Preparing for Compliance by 2 August 2026

From 2 August 2026, organisations will become subject to the transparency obligations set out in Article 50 of the EU AI Act (Regulation (EU) 2024/1689).

Article 50 introduces transparency requirements for providers and deployers in relation to certain AI system functionalities and use cases that may create transparency risks for individuals. Whilst much of the EU AI Act focusses on obligations on high-risk AI systems, Article 50 obligations may also apply to certain limited-risk systems. As a result, many organisations will need to implement governance, disclosure and content-labelling measures to ensure users are appropriately informed about the use of certain AI systems and AI-generated content.

Importantly, these obligations are generally not affected by the proposed implementation delays contained in the EU Digital Omnibus Proposal, except for the marking obligations contained in Article 50(2) of the EU AI Act, i.e. relating to AI systems that generate synthetic audio, image, video or text content and that were placed on the market before 2 August 2026. Per the provisional agreement reached between the European Council and European Parliament on 7 May 2026 (available here), the introduction of these obligations will be delayed until 2 December 2026.

Organisations should continue their compliance preparations notwithstanding the broader discussions regarding the timing of other EU AI Act requirements, albeit remain mindful of the impact of the phased implementation dates.

Article 50 Transparency Obligations

The principal transparency obligations under Article 50 are summarised below alongside non-exhaustive examples of use cases from the Commission’s draft Guidelines on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act (“Guidelines”).

Practical Challenges: Content Labelling and Disclosure

While the obligations under Article 50 are relatively straightforward, implementing the required disclosure and content-labelling measures may present practical challenges. Organisations will need to determine how AI-generated content is identified and how disclosures are incorporated into user interactions and content publication workflows.

Providers: Compliance is likely to require technical solutions capable of embedding machine-readable indicators into AI-generated content. Depending on the use case, this may include metadata tagging, watermarking, cryptographic provenance mechanisms or machine-readable audit logs.

Deployers: The European Commission has published a set of voluntary transparency icons designed to support compliance with Article 50(4) (i.e. the rules around disclosing the use of AI to create or manipulate deepfakes or public interest text). These are supplemented by detailed implementation guidance in the Code of Practice on Transparency of AI-Generated Content (which, at the date of writing, remains subject to an adequacy assessment by the European Commission and The AI Board).

Compliance is not limited to implementing technical content-labelling measures. Organisations must also ensure disclosures are clear, accessible and provided at the appropriate point in the user journey. This may require updates to onboarding processes, user interfaces and content publication workflows, as well as consideration of how Article 50 disclosures interact with existing transparency obligations under the GDPR and other regulatory regimes. In many cases, integrating AI-related disclosures into existing notices and transparency frameworks may be preferable to introducing standalone AI-specific statements.

Organisations should also consider where responsibility for compliance sits. Providers and deployers may each have distinct transparency obligations in relation to the same AI-enabled product or service, making it important to review contractual arrangements with AI vendors and service providers to ensure responsibilities for disclosures, content labelling and compliance monitoring are appropriately allocated.

Key Takeaways from the Draft Guidelines

The draft Guidelines provide valuable insight into how regulators are likely to interpret Article 50 in practice, although the final text may evolve following the consultation process. Key considerations include:

1. The Guidelines clarify that disclosure is not required where interaction with an AI system is sufficiently obvious. Whether this exemption applies is highly context-specific and depends on the characteristics and expectations of the relevant audience. Organisations should therefore exercise caution before concluding that disclosure is unnecessary and document the basis for any reliance on the exemption.
2. Accessibility is a key consideration. Disclosures should be clear, understandable and appropriately tailored to vulnerable individuals, including children.
3. Transparency measures should be built into the design of AI systems. Providers are expected to incorporate disclosure mechanisms into the way AI systems interact with users.
4. The Guidelines take a broad view of the content-labelling obligations. Content requiring marking may include AI-generated or AI-manipulated translations, summaries, image alterations and other modifications that materially affect the meaning or appearance of content.
5. The Commission adopts a broad interpretation of deepfakes. Organisations should not assume the obligation is limited to sophisticated impersonation content, as AI-generated or AI-manipulated content may fall within scope where it closely resembles a real person, object, place or event and has the potential to mislead.
6. The Guidelines confirm that agentic AI systems may fall within scope where their actions generate outputs intended to be directly perceived by users.

Recommendations for Organisations

The draft Guidelines suggest that regulators will focus not only on whether disclosures are provided, but also on whether they are clear, accessible and effective in the relevant context. Organisations should therefore approach Article 50 as both a technical and operational compliance exercise.

With the application date approaching, organisations should consider taking the following steps:

• Map relevant AI use cases: Identify AI systems that are provided, deployed or embedded within business processes and assess whether Article 50 transparency obligations are triggered.
• Review user interactions: Determine where users interact with AI systems and whether existing disclosures are sufficient.
• Assess content-generation workflows: Identify AI-generated or AI-manipulated outputs and evaluate whether marking and disclosure mechanisms are required.
• Develop labelling standards: Establish organisation-wide standards for AI disclosures, content labelling and user notifications.
• Update governance frameworks: Ensure AI governance policies, procurement processes and deployment reviews include Article 50 assessments.
• Document exemption analyses: Where relying on an exemption (such as the “obviousness” exemption), maintain records supporting that assessment.
• Review vendor arrangements: Ensure responsibility for Article 50 compliance is appropriately allocated between providers and deployers, particularly where third-party AI tools are used.

Conclusion

With less than two months until the majority of the transparency obligations take effect, organisations should ensure they understand which AI use cases fall within the scope of Article 50 and whether existing disclosure and content-labelling practices are sufficient. Early action may be important where technical, operational or contractual changes are needed to support compliance.

Trainee Solicitor Jennifer Petch also contributed to this blog post.

This post is as of the posting date stated above. Sidley Austin LLP assumes no duty to update this post or post about any subsequent developments having a bearing on this post.