Extract from Melina Efstathiou’s article, “Half In Force.”
There is a comfortable assumption circulating in legal and compliance circles right now and it goes something like this: The EU AI Act has been postponed, the pressure is off and 2 August 2026 no longer means what it used to mean. General Counsel have relaxed. Compliance programmes have been quietly deprioritised. Board papers that once flagged August as a hard deadline have softened their language.
That exhale is premature. What has actually happened to the EU AI Act’s timeline is considerably more precise, more conditional and more urgent than “postponed” — and the gap between the perception and the legal reality is exactly the kind of thing that turns into a governance failure nobody saw coming.
What is actually happening on 2 August 2026
Start with the baseline because it has not moved.
The AI Act entered into force on 1 August 2024, with a staggered timetable. Chapters I and II — general provisions and prohibited AI practices — have applied since 2 February 2025. Chapter V, covering general-purpose AI models, along with governance provisions and penalties, has applied since 2 August 2025. None of that has changed and none of it is affected by anything currently under discussion.
