Article 50 transparency obligations: the AI Act duty that has applied since 2 August 2026 and was not postponed

Regulation (EU) 2026/1744 fixes the core high-risk AI dates at 2 December 2027 for Annex III and 2 August 2028 for Annex I, but Article 50 has applied since 2 August 2026. Organisations deploying chatbots, generating synthetic content or producing deepfakes therefore need the relevant transparency measures in place now.

There is a quiet misreading travelling through compliance teams right now. The headlines about the Digital Omnibus focused on later high-risk dates. Article 50 is a separate transparency layer and has remained applicable since 2 August 2026. Its four duties are not a blanket label rule: the applicable action depends on the use case and on whether the organisation is provider or deployer.

This post is about that gap. It is the companion to our practical guide on Article 50 labelling and detection, which walks through the technical marking work itself. Here the focus is narrower and more urgent: what the Omnibus did and did not change, and why 2 August 2026 is still a live date.

What Article 50 actually requires

Article 50 sits outside the high-risk regime. Risk classification does not decide whether it applies, but neither does every use of AI trigger the same duty. The concrete use and the organisation's role as provider or deployer determine the result.

There are four core situations. Under paragraph 1, providers of systems intended to interact directly with people inform them that they are interacting with AI unless this is obvious from the context. Under paragraph 2, providers of systems generating synthetic audio, image, video or text machine-mark that output. Under paragraph 3, deployers inform people exposed to emotion recognition or biometric categorisation. Under paragraph 4, deployers disclose deepfakes and certain AI-generated or manipulated public-interest text, subject to its exceptions and modalities.

The reach can be wider than a high-risk inventory suggests, but scope still needs analysis. A marketing team drafting copy does not automatically receive the same duty as the provider of the model; a service desk may rely on a third-party provider's paragraph 1 design; and a media outlet may have separate provider or deployer duties depending on the content and publication workflow.

The one date that did move within Article 50

There is a single nuance worth getting right, because it is where the genuine relief lives. The general transparency duties under Article 50, including the chatbot disclosure and the deepfake disclosure, have applied since 2 August 2026. But the specific machine-readable marking obligation on providers of generative AI under Article 50(2) carries a separate, slightly later timing for content systems that were already on the market.

In practice, Article 50 has applied since 2 August 2026. Regulation (EU) 2026/1744 gives only providers of relevant paragraph 2 systems placed on the market before that date until 2 December 2026 for machine-readable marking. The transition does not automatically extend to deployer duties under paragraphs 3 and 4.

The deadlines side by side

The clearest way to see what the Omnibus did is to put the obligations in one timeline. The high-risk dates moved. Article 50 did not.

The pattern is deliberate. The Omnibus targeted the part of the AI Act that was hardest to operationalise on time, the high-risk classification regime, while leaving the transparency obligations untouched precisely because they are seen as low-burden and citizen-facing. For a fuller view of every milestone, see our overview of the AI Act deadlines for 2026, 2027 and 2028.

Why this catches organisations off guard

The trap is structural rather than careless. Most compliance programmes were built around a single anchor date of 2 August 2026, with high-risk classification as the centre of gravity. When the Omnibus moved that anchor, the natural response was to move the entire programme with it. But Article 50 was never bolted to the high-risk regime. It rode along on the same calendar date by coincidence, not by design, and now that the high-risk date has slipped, Article 50 has been left standing alone on 2 August 2026.

There is a second reason it gets missed. Providers carry the paragraph 1 interaction disclosure and paragraph 2 machine-marking duty; deployers carry the paragraph 3 information duty and paragraph 4 disclosures. An organisation can hold different roles for different systems, but integrating or using a third-party model does not by itself make it the provider.

Practical steps now that Article 50 applies

The work is narrower than the high-risk programme, which is good news given the timeline. Start by mapping every point where your organisation touches the four Article 50 situations: customer-facing chatbots and assistants, any generative model that produces public content, any deepfake or synthetic media production, and any AI-assisted text published to inform the public. This is a use-case inventory, not a system inventory, and it usually surfaces deployments that the high-risk mapping never caught.

For direct AI interaction, confirm that the provider has designed the required disclosure and determine whether your own branding or modification changes your role. For synthetic output, ask the provider how paragraph 2 marking is delivered and preserved. Separately assess whether your use creates a paragraph 4 deepfake or public-interest-text disclosure. Do not collapse provider marking and deployer disclosure into one generic label.

The supporting Code of Practice on transparency of AI-generated content is still being finalised, with the Commission having run consultation rounds through the first half of 2026. The Code is voluntary, but adherence will be the practical evidence that your marking approach is adequate. Building toward its layered marking expectations now is more efficient than retrofitting later. Do not wait for the final text to start the engineering work, because the obligation itself does not depend on the Code being finished.

Where this fits in the wider picture

It is worth stepping back. The AI Act already has obligations in force today: the AI literacy duty under Article 4 has applied since 2 February 2025, and the prohibited practices since the same date. Article 50 is the next hard edge on the calendar, arriving on 2 August 2026, ahead of the postponed high-risk regime in late 2027. Organisations that read the Omnibus as a general reprieve have the sequence backwards. The obligations that touch the most people, transparency and literacy, are the ones that are live or arriving first. The heavy, system-specific high-risk obligations are the ones that have been pushed out.

If your AI governance roadmap currently has everything clustered around late 2027, separate the Article 50 work immediately and put it where it belongs: in force since 2 August 2026, not at the later high-risk dates.

Want to know where your organisation stands? The free AI transparency scan by Embed AI checks your position on AI disclosure, content marking, deepfakes, policy and evidence in seven questions, with an instant score and your biggest gap.