The European Parliament in June 2026 approved final amendments to Regulation (EU) 2024/1689, the Artificial Intelligence Act, as part of the Digital Omnibus simplification package. The vote concludes the trilogue process: a political agreement between the Parliament and the Council was reached on 7 May 2026. The amended text proceeds to formal Council endorsement before publication in the Official Journal of the European Union.
The amendments revise the classification provisions governing high-risk AI systems under Annex III of the Act. For systems listed directly in Annex III - including those used in biometric identification, critical infrastructure management, education, employment screening, access to essential services, law enforcement, migration management, and administration of justice - the application date shifts to 2 December 2027. For AI systems embedded in products subject to existing EU sectoral safety legislation, the application date moves to 2 August 2028. A new standalone prohibition bans the placement on the market of AI systems that generate sexual or intimate images of identifiable natural persons without their consent, extending the category of prohibited practices under Article 5 of the Act.
Compliance programmes for Annex III deployers - including banks, insurers, healthcare providers, employment screening platforms, recruitment agencies, credit-scoring services, and law enforcement bodies - must be revised to reflect the new deadlines. Operators whose AI systems were calibrated to the original August 2026 or August 2027 timelines gain additional preparation time but remain subject to the governance and transparency provisions already in effect. Providers and deployers of AI systems that generate intimate imagery must take immediate withdrawal action: the new prohibition does not carry a deferred application date.
The Council must formally adopt the amended text before it enters into force, and the date of publication in the Official Journal determines the start of the revised compliance clocks. The general-purpose AI model obligations under Chapter V of the Act and the Artificial Intelligence Office's supervisory mandate are not altered by the Digital Omnibus amendments and continue to apply on the original schedule. Implementing acts and delegated acts under the Act remain in preparation, and the Artificial Intelligence Board's guidance documents are not yet finalised.
Licentium advises clients on EU Artificial Intelligence Act compliance strategy, Annex III high-risk classification assessments, and GPAI model obligations. We assist technology companies, financial institutions, and healthcare providers with compliance programme design and gap analysis against the revised timelines. Work we undertake includes high-risk AI system classification, internal governance documentation, prohibited practice assessments, vendor due diligence support, and regulatory horizon monitoring across EU member states.