The delay becomes law this month. This still bites on 2 August.

This analysis first appeared as the weekly Praxikon newsletter.

<h2>The AI Act converges in two moves this month</h2> <p>In just over three weeks, on <strong>2 August 2026</strong>, two parts of the AI Act become genuinely enforceable. And shortly before that, expected between 18 and 30 July, the high-risk delay is due to appear in the Official Journal. For the first time, the picture is clear enough to plan against. Start with what bites on 2 August, because none of it has been postponed:</p> <ul> <li><strong>Article 50 (transparency).</strong> A chatbot on your site or client portal must make clear that the user is talking to AI. AI-generated text, images or audio you publish must be recognisable as AI-made. Content that a human has reviewed, with someone taking editorial responsibility, falls outside this. For marking content already on the market, a transition period runs until 2 December 2026.</li> <li><strong>GPAI model obligations.</strong> From the same date, enforcement becomes sharp, with fines up to 3% of global turnover or 15 million euro for model providers.</li> </ul> <h2>And this has shifted by a year</h2> <p>The European Parliament (16 June) and the Council (29 June) have formally approved the Digital Omnibus. This moves the high-risk obligations for Annex III systems (recruitment and selection, credit scoring, education, essential services) to <strong>2 December 2027</strong>, and systems embedded in regulated products to 2 August 2028. The text is expected in the Official Journal this month and enters into force three days later. Note: until that publication, the original text still applies legally. The delay is all but done, but not yet formally law.</p> <p>Our advice is unchanged: use the extra time to build your AI register and evidence file, not to wait. The registration and documentation burden under the high-risk rules is substantial, and the organisations that want to be ready on 2 December 2027 are starting now.</p> <h2>Action for this week</h2> <p>Map where in your organisation AI talks directly to people or publishes content: chatbots on your site or client portal, generated client emails, marketing visuals, voice assistants. That list is your Article 50 scope. Per item, the question is simple: is there a clear disclosure, or does a human take final editorial responsibility? Anything with neither is work that must be done before 2 August.</p> <p>If you want to tackle this with your team in a structured way, including practice scenarios and evidence per employee: see the <a href="https://learnwize.ai/article-50-transparency-sprint?utm_source=praxikon&utm_medium=referral&utm_campaign=praxikon_weekly&utm_content=editie-2026-07-14">Article 50 Transparency Sprint by LearnWize</a>.</p> <p>You receive this weekly AI Act update because you signed up with Praxikon or used one of our tools or templates. You can unsubscribe with one click below.</p> <p>See you next Tuesday.<br/>Zahed Ashkara<br/>Praxikon</p>