EU AI Act Amendments Explained: Parliament Approves Key Changes

The European Parliament has given its strong endorsement to significant amendments to the EU AI Act, with a substantial majority of 423 Members of European Parliament (MEPs) voting in favor, 57 against, and 174 abstentions. This crucial vote on June 16, 2026, follows a provisional agreement that was reached on May 7 between the Council of the European Union and the lead MEP negotiators, solidifying the direction of future AI regulation within the bloc. Although the amended regulation has not yet been formally adopted and published, its passage through the Parliament signals that the changes are imminent. Businesses are therefore urged to proactively assess and adapt their AI governance and compliance frameworks. The current article, the first in a two-part series, offers a detailed explanation of the ten most impactful amendments to the Act, providing essential guidance for enterprise practitioners to prepare for the updated regulatory landscape. The amendments stem from proposals made by the European Commission in November 2025 as part of its Digital Omnibus package. The stated objectives were to simplify and streamline key elements of the law, reduce compliance burdens for businesses, and foster European competitiveness in the AI sector. Despite some adjustments to timelines and postponements, the fundamental goals, logic, and structural integrity of the EU AI Act are largely preserved. This indicates that the EU remains steadfast in its commitment to stringent and comprehensive AI regulation. One notable change highlighted is the prohibition against using AI systems to generate or manipulate non-consensual intimate imagery or child sexual abuse material (CSAM). This prohibition will be applicable from December 2, 2026, and mandates that providers of general-purpose AI (GPAI) systems implement robust, state-of-the-art guardrails and technical safeguards to mitigate such risks. Additionally, the article clarifies the updated compliance timelines for high-risk AI systems. While some parts of the Act, such as prohibited AI practices and AI literacy obligations, are already in force, the compliance date for high-risk AI systems listed in Annex III (e.g., education, employment, law enforcement) is now August 2, 2026, and for those integrated into regulated products (Annex I), it is August 2, 2027.