The countdown to Article 50 compliance has entered its final days. On 20 July 2026, the European Commission published its final guidance on the EU AI Act’s transparency obligations, giving providers and deployers a clearer interpretation of what regulators expect before the rules begin applying on 2 August. The guidance covers AI interaction disclosures, AI-generated and manipulated content, emotion recognition and biometric categorisation systems, and deepfakes.
For providers and deployers, the timeline is now measured in days rather than months. The transparency obligations under Article 50 of the EU AI Act become directly applicable on 2 August 2026. Non-compliance can result in administrative fines of up to €15 million or 3% of worldwide annual turnover, whichever is higher, under the AI Act’s enforcement framework.t
One important qualification applies to only one of the four transparency obligations. General-purpose AI systems already placed on the EU market before 2 August 2026 have until 2 December 2026 to comply with the machine-readable marking and detection requirements under Article 50(2). By contrast, the obligations relating to AI interaction disclosures, emotion recognition and biometric categorisation disclosures, and deepfake transparency apply from 2 August 2026, with no equivalent transitional period for systems already on the market.
The Commission’s guidelines are non-binding interpretative guidance adopted under Article 96 of the AI Act. They explain how the Commission expects providers and deployers to implement the transparency obligations in practice but do not create new legal obligations beyond those established by the Regulation itself. Alongside the guidelines, the Commission has published the Code of Practice on Transparency of AI-Generated Content. The Code is voluntary and has been recognised by the Commission and the European Artificial Intelligence Board as an adequate means of demonstrating compliance with Article 50, while organisations remain free to demonstrate compliance through other appropriate methods.
For compliance teams, this is no longer a planning exercise. The remaining days before 2 August should be used to verify implementation rather than draft new policies. Review user-facing disclosures, AI-generated content workflows, machine-readable marking where applicable, supplier documentation, and contractual responsibilities to ensure deployed systems reflect both the legal requirements of Article 50 and the Commission’s implementation guidance. Organisations that can demonstrate operational compliance from day one will be better positioned if regulators request evidence after the obligations take effect.
With the publication of the final guidelines, the Commission has reduced much of the uncertainty surrounding Article 50 implementation. For providers and deployers, the focus now shifts from interpreting the Regulation to demonstrating compliance in practice. Before 2 August 2026, organisations should verify that disclosure mechanisms, technical controls, governance documentation, and supplier arrangements align with both the legal requirements of the AI Act and the Commission’s guidance. Those that can evidence compliance from day one will be better prepared for regulatory scrutiny as enforcement begins.