EU AI Act · Myth-buster
Short version: no, the whole EU AI Act was not delayed. The part that hits online sellers and creators — Article 50 transparency (labelling AI content, chatbot and deepfake disclosure) — still applies from 2 August 2026.
If you've seen headlines saying "the EU AI Act has been pushed back," you're not imagining it. In June 2026 the EU agreed a package called the Digital Omnibus, and a lot of coverage framed it as "the AI Act is delayed." That's the wrong takeaway for most sellers. The delay applies to high-risk AI systems — not to the transparency rules that cover AI-generated images, chatbots and deepfakes. Those still land on schedule.
The Council of the EU gave the Digital Omnibus its final green light on 29 June 2026. It's a simplification package, and its headline move was to push back the deadlines for high-risk AI systems (the Annex III category — think AI used in hiring, credit scoring, medical devices, critical infrastructure). Those deadlines moved to:
If you're a shop owner using AI to write product descriptions or generate listing images, none of that is you. High-risk rules were never your deadline, so the "delay" everyone's talking about doesn't move anything on your plate.
Here's the part that didn't move. Article 50 transparency obligations still apply from 2 August 2026. The Omnibus left them alone. If you use AI to make content for the public — or run a chatbot — this is the date that matters to you.
The one narrow exception: for generative-AI systems placed on the market before 2 August 2026, the machine-readable marking requirement (Article 50(2), the "invisible" watermark) gets a short grace period to 2 December 2026. The visible, human-facing disclosures still start 2 August 2026.
In plain terms, if any of these describe you, Article 50 is in scope from 2 August 2026:
None of this requires a lawyer or a big budget. Mostly it's a label and a line of disclosure in the right places.
For Article 50 breaches, the statutory maximum is up to €15 million or 3% of global annual turnover, whichever is higher (AI Act Article 99). That's a legal ceiling, not a prediction — nobody is forecasting a €15M fine for a mislabelled listing image. Ignore anyone quoting €35M or 7% here: that higher tier is for prohibited AI practices, not transparency rules. The reason the ceiling matters isn't the worst case; it's that a fix takes about ten minutes and the exposure doesn't.
The rule is Regulation (EU) 2024/1689, Article 50 — the EU AI Act. Read the official text on EUR-Lex: Regulation (EU) 2024/1689 →. The Digital Omnibus is the June 2026 amending package that deferred the high-risk deadlines above.
or see every compliance deadline, or read the full EU AI Act guide.
Not the whole thing. The Digital Omnibus (final Council approval 29 June 2026) delayed the high-risk AI deadlines to 2 December 2027 and 2 August 2028. The Article 50 transparency rules — labelling AI content, chatbot and deepfake disclosure — were not delayed and still apply from 2 August 2026.
Yes. Article 50 transparency obligations apply from 2 August 2026. The only tweak is a grace period to 2 December 2026 for machine-readable marking on generative-AI systems already on the market before 2 August 2026. Visible disclosures start on schedule.
Up to €15 million or 3% of global annual turnover, whichever is higher (AI Act Article 99) — a statutory ceiling, not a likely outcome. The higher €35M / 7% tier applies only to prohibited AI practices, not to transparency rules.
Informational only, not legal advice, and not affiliated with the European Commission. Last reviewed 2026-07-04.