Written by: Haim Ravia, Dotan Hammer
On March 18, 2026, the European Parliament’s Committee on the Internal Market and Consumer Protection (IMCO) and the Committee on Civil Liberties, Justice and Home Affairs (LIBE) adopted their joint negotiating position on the European Commission’s proposed Digital Omnibus on AI, which seeks to simplify the implementation of the EU AI Act. The Final Compromise Amendments will serve as the Parliament’s mandate for trilogue negotiations with the Council and Commission—a process that will determine the final shape of the amendments to the AI Act.
A central point of contention has been the treatment of AI literacy obligations under Article 4 of the AI Act. The Commission had proposed converting these from mandatory requirements into voluntary measures merely “encouraged” by Member States. The Parliament’s compromise amendments push back: they retain the obligation for providers and deployers of AI systems to take measures to ensure a sufficient level of AI literacy among their staff and other persons dealing with the operation and use of AI systems on their behalf. However, the Parliament added important qualifications: the obligation does not guarantee a specific level of AI literacy for any individual, and the Commission is instructed to issue guidance on the practical implementation of this obligation. In a parallel provision, Member States and the Commission are encouraged to support AI literacy in society and among the general population, and to facilitate the efforts of providers and deployers—particularly SMEs—in fulfilling their obligations.
On the contentious question of timeline delays for high-risk AI systems, the Parliament adopted a conditional mechanism that differs from the Commission’s proposed blanket postponement. Under the compromise, the high-risk system rules under Chapter III will apply following a Commission decision confirming that adequate compliance support measures are available—six months after that decision for Annex III systems and twelve months for Annex I systems. However, if the Commission does not issue such a decision, hard backstop dates apply: December 2, 2027, for Annex III systems and August 2, 2028, for Annex I systems. This approach seeks to balance the acknowledged lack of harmonized standards and delays in designating national competent authorities with the need to prevent indefinite postponement of fundamental rights protections.
The compromise amendments also address the establishment of EU-level regulatory sandboxes by the AI Office. The Parliament insists on robust safeguards: where innovative AI systems involve the processing of personal data or otherwise fall under the supervisory remit of national authorities, national data protection authorities and the EDPB must be associated with the sandbox’s operation and involved in supervision to the extent of their respective tasks and powers, in accordance with the GDPR and related regulations. The Parliament further specifies that sandboxes should provide priority access to SMEs, including startups. Rules on real-world testing of high-risk AI systems outside sandboxes were also refined, including provisions for new voluntary real-world testing agreements between Member States and the Commission for AI-enabled products under Union harmonization legislation.
Click here to read the European Parliament’s Final Compromise Amendments on the Digital Omnibus on AI.
