What came into force with the EU’s AI Act this week – and what didn’t

What the report says

The European Union has brought the next stage of its Artificial Intelligence Act into force, in a move Al Jazeera describes as the first comprehensive AI law of its kind. The new step, which took effect on August 2, adds transparency rules for AI systems that interact with people or generate synthetic content. It also requires disclosure when emotion-recognition or biometric categorisation tools are used, with some law-enforcement exceptions. The Act does not ban these systems, but companies that fail to comply could face significant fines.

Al Jazeera reports that the immediate effect is mainly operational for businesses, which now need to identify where AI is embedded in products, customer services and internal processes, including third-party tools, and then adjust compliance procedures. The rules are meant to sit alongside existing digital laws rather than replace them, similar to how the EU’s privacy regime works with the GDPR.

The article also says the Act’s more demanding “high-risk” obligations, covering areas such as biometrics, employment, education, essential services and migration, asylum and border management, were postponed in May until December 2, 2027. EU officials have framed that delay as a practical implementation step while technical standards and guidance are completed. Critics, including digital rights and racial justice advocates, argue the postponement weakens protections and leaves people affected by automated decisions, especially migrants and asylum seekers, exposed for longer.

The law is being phased in over time: earlier rules on prohibited practices and AI literacy started in February 2025, general-purpose AI model rules followed in August 2025, and the newest transparency provisions now apply.