AI Regulation Tracker  /  Transparency deadline

The EU AI Act Deadline That Did Not Move

The Digital Omnibus pushed the headline high-risk rules to 2027. The Article 50 transparency duties still take effect on August 2, 2026, and they reach professionals in the United States whose AI output lands in Europe.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 3, 2026  ·  Last updated July 3, 2026

The delay everyone noticed, and the deadline they missed

For most of the spring, the story out of Brussels was relief. The European Commission's Digital Omnibus, a simplification package aimed at the EU AI Act, spent months moving through negotiation, and on June 29, 2026 the Council of the EU gave it final approval. The headline that traveled was simple. The hardest part of the AI Act, the high-risk rules, would not bite in August 2026 after all. Companies exhaled.

Read the fine print and a different fact appears. The Omnibus moved the high-risk timelines. It left the transparency obligations in Article 50 almost exactly where they were. Those obligations still apply from August 2, 2026. As the law firm Gibson Dunn put it in its client alert, "The Article 50 transparency obligations for AI systems largely remain on the original schedule."

So the deadline that everyone stopped worrying about is roughly thirty days away, and it is the one most working professionals are actually exposed to.

What Article 50 actually requires

Article 50 is the AI Act's transparency chapter. It is short, and it is concrete. Four duties matter for most professionals.

• Tell people they are talking to an AI. Providers must design systems that interact directly with people so that a person is informed they are dealing with an AI, unless that is already obvious to a reasonably observant person.
• Mark AI-generated content so a machine can read it. Providers of systems that generate synthetic audio, image, video, or text must mark the output in a machine-readable format so it is detectable as artificially generated or manipulated.
• Disclose deepfakes and AI-written public-interest text. Deployers who use AI to generate or manipulate image, audio, or video content that is a deep fake must disclose that it was artificially generated. A parallel duty applies to AI-generated text published to inform the public on matters of public interest.
• Give notice for emotion recognition and biometric categorization. Deployers of those systems must tell the people exposed to them that the system is operating.

None of this requires a compliance department. It requires knowing where AI sits in your workflow and being honest about it in a form that both a human and a detection tool can see.

Yes, it can reach you in the United States

The reflex for a US professional is to assume an EU rule stops at the EU border. Article 2 of the AI Act says otherwise. The Regulation applies to providers placing AI systems on the EU market irrespective of whether they are established in the Union or in a third country, and it applies to providers and deployers located in a third country "where the output produced by the AI system is used in the Union."

Translate that into practice. A US firm running a client-facing AI chatbot that EU residents can use, a US marketing team publishing AI-generated video that circulates in Europe, or a US professional sending AI-drafted public content that reaches EU readers can all be captured. The trigger is where the output lands, not where you sit.

The narrow grace period, and what it does not cover

There is one carve-out worth knowing, and it is smaller than it sounds. AI systems already placed on the market before August 2, 2026 get a short grace period, until December 2, 2026, to meet the machine-readable marking duty in Article 50(2). That is a four-month runway for the watermarking piece on existing tools. It is not a delay of Article 50 as a whole, and it does not touch the duty to tell people they are interacting with an AI or to disclose deepfakes. Treat it as breathing room for one technical task, not as a reason to wait.

What did move: high-risk to 2027

The Omnibus did real work on the high-risk regime, and it is worth being precise so you do not overcorrect. High-risk obligations for stand-alone Annex III systems, the use-based category that covers areas like employment, credit, and access to essential services, are deferred to December 2, 2027. High-risk obligations for AI embedded in regulated products under Annex I, such as medical devices and machinery, are deferred to August 2, 2028. If your exposure is genuinely high-risk in the AI Act sense, you gained more than a year. If your exposure is transparency, you gained nothing.

The penalty math

Article 99 sets the fines, and Article 50 sits in the middle tier. Breaches of the transparency obligations can draw penalties of up to 15 million euros or 3 percent of total worldwide annual turnover, whichever is higher. The top tier, up to 35 million euros or 7 percent, is reserved for the prohibited practices in Article 5. The point is not that a small firm will be fined 15 million euros on August 3. It is that transparency is a supervised obligation with a real penalty attached, not a best-practice suggestion.

A compliance pass you can run this month

You do not need a consultant to get most of the way there. Five steps cover the common cases.

• Inventory your AI touchpoints. List every place AI talks to a person or produces content that leaves your office: chatbots, intake assistants, AI-generated images and video, AI-drafted public posts.
• Add the interaction disclosure. Where AI interacts with people, add a plain line that says so. "You are chatting with an AI assistant" is enough when it is clear and visible.
• Turn on content marking. For synthetic media, enable the provenance and watermarking features your tools offer so the output is detectable as AI-generated. The Commission published a Code of Practice on marking AI-generated content on June 10, 2026 to guide exactly this.
• Label deepfakes and public-interest text. If you publish AI-manipulated video or AI-written content meant to inform the public, disclose it in the content itself.
• Write it down. Keep a one-page record of what you disclose and where. If a question ever comes, the firm that can show its map looks careful. The firm that cannot looks exposed.

The lesson of this deadline is smaller and more useful than the headline suggested. The scary part of the AI Act slid to 2027. The part that touches ordinary professional work did not. August 2 is the date to keep.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.