The Legal edition
The AI news of that week, in plain language.
This week is all about rules and responsibility: the AI Act comes into force, deepfake fraud is growing, and there are smart tools for research and documents.
EU AI Act comes into force: transparency and AI knowledge now required, high-risk postponed
The big deadline for recruitment and selection software is moving back, but two duties do start as planned on 2 August 2026.
The Digital Omnibus, a change to the EU AI Act, was published on 24 July 2026 and came into force on 27 July. The deadline for high-risk AI (such as recruitment and credit systems) moves from 2 August 2026 to 2 December 2027. But two duties do start as planned on 2 August 2026: the transparency duty (Article 50) and the AI literacy duty for staff (Article 4). Dutch supervisors (AP and RDI) will enforce this. Fines run up to 35 million euros or 7% of worldwide annual turnover.
Background for subscribersDo not rely on the whole postponement message. The transparency duty (Article 50) and the AI literacy duty (Article 4) start as planned on 2 August 2026. For the first wave of enforcement, HR, education and government are mainly in view. Employers must inform staff and the works council in advance for systems that affect promotion, dismissal or task allocation. Only the deadline for high-risk AI, such as CV ranking software, has moved to 2 December 2027. Medical AI largely falls under a later deadline.
AI recruitment stays high-risk: human oversight required
78% of Dutch recruiters use AI tools, which saves 6.5 hours a week on average. Even so, CV ranking software falls under the high-risk category of the EU AI Act: those systems require documentation, human oversight and information for candidates. An internal HR chatbot for frequently asked questions does not fall under it.
Background for subscribersThis is a taster.
Subscribers read the whole Legal edition: every story, each with its background in plain language.
Subscribe →