The EU published its AI-content transparency Code, and set a 22 July signatory deadline before Article 50 bites

The 2 August 2026 start of the EU AI Act Article 50 transparency rules has a companion most planning notes miss. On 10 June 2026 the European Commission published a voluntary Code of Practice on transparency of AI-generated content, and the AI Office set 18:00 CET on 22 July 2026 as the cut-off to be named on the first list of signatories. That list is due out before the obligations apply. So there are really three dates to keep straight, and the nearest one is only about three weeks away.

What the Code is, and what it is not

The Code was drawn up by independent experts in a multi-stakeholder process facilitated by the AI Office. It is a practical playbook for the marking and labelling side of Article 50: it helps a provider or deployer demonstrate compliance with Article 50(2) (machine-readable marking of synthetic audio, image, video and text), Article 50(4) (labelling deepfakes and certain AI-generated public-interest text), and Article 50(5) (how and when that information is delivered). Signing lets you point to its agreed measures as evidence you are meeting those duties.

What it is not is a new obligation. Adherence to the Code is voluntary, but the Article 50 transparency requirements it maps to are binding law whether you sign or not. Signing does not add duties; it gives you a ready-made way to show you are discharging the ones you already have, with more legal certainty across Member States. Not signing is fine too, as long as you can show compliance some other way.

Why 22 July matters, and why it does not

To be listed among the initial signatories that the AI Office publishes before 2 August 2026, the completed signatory form has to be in by 18:00 CET on 22 July 2026. Missing that window does not lock you out: you can still sign later, and you can still comply without signing at all. The deadline is about being on the first published list, which is a credibility signal to customers and regulators, not a compliance cliff. Treat it as a reason to decide your position now rather than a date that changes what you owe.

The one Article 50 date that did move

The digital omnibus that reshaped the surrounding AI Act deadlines cleared its final votes this summer: the European Parliament adopted it on 16 June 2026 and the Council of the EU gave its green light on 29 June 2026, with publication in the Official Journal still to come. It left the general Article 50 obligations on 2 August 2026, but it added one grace period worth knowing. For AI systems already placed on the market before 2 August 2026, the Article 50(2) machine-readable marking obligation is delayed to 2 December 2026. Systems launched on or after 2 August 2026 get no such runway, and the interaction-disclosure, deepfake-labelling and delivery duties all still start on 2 August.

What is at stake if you get it wrong

From 2 August 2026, national authorities can enforce Article 50, and a breach sits in the tier that carries a fine of up to 15 million euros or 3 percent of worldwide annual turnover, whichever is higher. The transparency duties are also the ones most ordinary businesses meet first, well before any high-risk classification question, because they attach to everyday uses like a customer chatbot or AI-generated marketing copy rather than to a narrow list of regulated systems.

The disclosure itself is not the hard part. Our free transparency-notice generator writes one tied to the exact Article 50 categories, so you can paste it into a chatbot, a footer or a content credit. If you are still not sure whether the rules reach you, our companion guide on whether your chatbot needs an AI disclosure walks the four Article 50 duties and the provider-versus-deployer split.