Commission Opinion of 8 July 2026 on the assessment of the Code of Practice on Transparency of AI-generated Content — C(2026) 4839 final

facilitates the effective implementation of the transparency obligations laid down in Article 50(2) and (5) AI Act.

(28) In light of the above, the Commission considers that Section 1 of the Code adequately facilitates the effective implementation of the transparency obligations laid down in Articles 50(2) and (5) AI Act and that providers of AI systems may rely on the Commitments and Measures set out therein to demonstrate compliance with those obligations.

3.3. Contribution to the effective implementation of Article 50(4) of the AI Act

(29) Section 2 of the Code aims to contribute to the effective implementation of the transparency obligations applicable to deployers of generative AI systems laid down in Articles 50(4) and (5) of the AI Act.

(30) Article 50(4) AI Act set out two transparency obligations for deployers of certain generative AI systems and their content:

(31) First, deployers of AI systems that generate or manipulate image, audio or video content constituting a deep fake within the meaning of Article 3(60) AI Act must disclose that the content has been artificially generated or manipulated. Where the deep fake content forms part of an evidently artistic, creative, satirical, fictional or analogous work or programme, the transparency obligation is limited to disclosure of the existence of such generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work.

(32) Second, deployers of AI systems that generate or manipulate text which is published with the purpose of informing the public on matters of public interest must disclose that the text has been artificially generated or manipulated. This obligation does not apply where the AI-generated or manipulated content has undergone a process of human review or editorial control, and where a natural or legal person holds editorial responsibility for the publication.

(33) Article 50(5) AI Act further specifies that the information referred to under Article 50(4) AI Act (i.e. the applied measure that discloses that content has been artificially generated or manipulated) should be provided to the natural persons exposed to the content in a clear and distinguishable manner at the latest at the time of first exposure to the content.

(34) The Commitments and measures set out in Section 2 of the Code cover each of the aforementioned elements of the obligation laid down in Articles 50(4) and (5) AI Act and facilitate their effective implementation by deployers of generative AI systems as follows.

(35) Commitment 1 facilitates the consistent and effective disclosure of the artificial origin of deep fakes or published text on matters of public interest. It does so, in particular, by providing (i) three common EU icons (provided in Annex 1 of Section 2) that Signatories may use to ensure disclosure and (ii) common design specifications on which Signatories not using the common EU icon should rely, while allowing for certain degree of flexibility. These design specifications (Measure 1.1) are further complemented with placement specifications (Measure 1.2) that ensure clear and distinguishable placement of the icons or labels used by the Signatories, at the latest at

EN 7 EN

the time of the exposure of the natural persons to the content, and that facilitate the effective implementation of Article 50(5) AI Act. Both sets of specifications rely on whether visual disclosure is possible. This ensures that the specifications are flexible and can be implemented across a range of use cases, sectors, and types of modalities, facilitating effective compliance by Signatories with Articles 50 (4) and (5) AI Act.

(36) Commitment 2 facilitates the effective implementation of the disclosure obligations laid down in Article 50(4) AI Act by ensuring that Signatories put in place or maintain internal processes, awareness measures, and review mechanisms to that end. The implementation of the measures can rely on existing processes and documentation and take into account the size and available resources of Signatories. This ensures proportionality and simplification, without necessarily affecting effectiveness. Specifically, Measure 2.1 ensures that Signatories put in place or maintain appropriate internal compliance processes and documentation that specify how they implement the disclosure obligations. For Signatories that regularly use AI systems to create deep fakes or published text on matters of public interest, this is complemented with a commitment for a process to ensure and prevent potentially non-labelled or mislabelled content. Measures 2.2 (Awareness and literacy) and 2.3 (Review, feedback and cooperation with authorities) further ensure the effective implementation of Article 50(4) AI Act with commitments by Signatories to raise awareness and literacy among relevant personnel, to review observed or reported mislabelled content, and to cooperate with competent authorities.

(37) Commitment 3 effectively enables deployers of generative AI systems to disclose deep fakes that form part of evidently artistic, creative, satirical, fictional or analogous work or programmes in a way that does not hamper the display or enjoyment of the work, including its normal exploitation and use, while maintaining the utility and quality of the work. In particular, it does so by allowing contextual disclosure where the information can be provided through the broader informative context, rather than as part of the content itself, while still mandating compliance with the design specifications of Measure 1.1 and the minimum requirements stemming from Article 50(5) AI Act.

(38) Commitment 4 facilitates the effective implementation of the exception to the transparency obligation under Article 50(4), subparagraph 2, AI Act. By making a distinction between media service providers as defined in Article 2(2) of Regulation 2024/1083 and other actors with respect to the measures to be taken, a pertinent distinction is made that does not impact the effectiveness of the exception. On the contrary, this distinction and the related expected measures ensure that the exception can only be invoked by Signatories that have taken appropriate practical measures supporting effective human review or editorial control, while also identifying a natural or legal person assuming editorial responsibility.

(39) Section 2 further contributes to the effective implementation of Article 50(4) AI Act by optional measures under the Code that are not legally required, but voluntary and nevertheless recommended. These measures are conducive to achieving the goal pursued by Article 50(4) AI Act to enable natural persons to effectively distinguish AI- generated or manipulated deep fakes and published text on matters of public interest. For example, Section 2 enables Signatories and the associations of which they are a part to jointly establish common good practices for their sector or industry, which can provide further details for the implementation of the Commitments and Measures specified in Section 2. Sub-measure 1.2.1 encourages Signatories to collaborate, on a

EN 8 EN

best effort basis, with actors whose services or products they use to further distribute or disseminate the deep fake or published text to preserve the icon or equivalent label with the objective of ensuring that the required disclosure accompanies the deep fake or published text throughout its further distribution. This directly serves the objective of the Code of Practice, mentioned by Recital 135 AI Act to facilitate cooperation with other actors along the value chain disseminating content and to enable the public to effectively distinguish AI-generated content from other content. Moreover, Measure 1.3 encourages Signatories to support the work of a task-force that is tasked, inter alia, with further developing and promoting the uptake of the common EU icon and providing a forum among Signatories to exchange good practices. Measure 2.1 encourages Signatories to publish information on their disclosure solution in order to enhance public understanding of their practices.