BRUSSELS, May 7 (Xinhua) -- The European Union (EU) on Thursday reached a political agreement on the new "digital omnibus on artificial intelligence (AI)." This is a major step, it said, toward simplifying the implementation of the AI Act, while protecting safety, fundamental rights, and public trust.
The AI Act was approved by the EU lawmakers in 2024, with most of its provisions set to take full effect on Aug. 2, 2026, giving governments and companies time to adapt to the new regulation framework. In response to concerns about excessive regulatory burdens, the European Commission proposed changes to the AI Act on Nov. 19, 2025, aimed at reducing compliance requirements for businesses.
The Thursday agreement introduces a more practical and phased approach to AI regulation, particularly for high-risk systems. Under the new timeline, AI systems used in sensitive sectors -- including biometrics, critical infrastructure, education, employment, migration, asylum, and border control -- will be subject to the AI Act starting Dec. 2, 2027. Meanwhile, AI embedded in products such as lifts and toys will follow from Aug. 2, 2028.
The phased rollout is intended to allow time for technical standards, compliance tools, and implementation guidance to be put in place before enforcement begins.
The agreement also strengthens safeguards against harmful AI applications by explicitly banning systems that generate non-consensual intimate or sexually explicit content. These include AI "nudification" apps and AI-generated child sexual abuse material.
"Our businesses and citizens want two things from AI rules. They want to be able to innovate and feel safe. Today's agreement does both," said EU tech chief Henna Virkkunen, adding that the EU will ensure that tools "for a smooth implementation of the AI Act are ready" to support companies in Europe.
The agreement still requires formal approval from the European Parliament and the Council of the EU. Once adopted, the amendments will be published and enter into force three days later. ■
