AI Act Under Pressure: Parliament and Council Negotiate Postponement of High-Risk Systems Rules in Digital Omnibus Deal
The European Parliament and the Council of the European Union are in late-stage negotiations to reach a deal on the so-called digital omnibus proposal — a package that would postpone certain AI Act rules on high-risk systems and address several adjacent digital policy issues, including a possible ban on nudification apps. The negotiations, which intensified in early May 2026, mark the first significant amendment to the AI Act since its adoption in 2024.
Why the postponement
The AI Act establishes a tiered risk framework: minimal-risk, limited-risk, high-risk, and prohibited systems. The high-risk systems category — covering AI used in critical infrastructure, education, employment, law enforcement, migration management, justice and democratic processes — was scheduled to become enforceable in August 2026. Industry has been lobbying for a one- to two-year postponement, citing technical challenges in implementing harmonised standards, the absence of finalised conformity assessment procedures, and the burden on SMEs developing AI systems.
The Commission position
The European Commission has expressed cautious openness to a limited postponement of 12 months for specific high-risk categories — particularly those where harmonised standards developed by CEN-CENELEC are not yet ready. The Commission insists that prohibited practices (social scoring, real-time biometric surveillance with limited exceptions, manipulative AI) remain subject to the August 2026 deadline without postponement. The position is presented as “calibrated postponement, not retreat.”
Civil society resistance
Digital rights organisations including European Digital Rights (EDRi), Access Now and Algorithm Watch have strongly opposed any postponement, arguing that delays would expose European citizens to unaccountable AI systems in critical contexts — particularly migration and law enforcement, where AI-driven decisions already affect fundamental rights. They warn against a creeping erosion of the AI Act under pressure from US tech companies and the second Trump administration’s deregulation agenda.
The nudification apps ban
Linked to the omnibus discussions is a parallel debate on banning so-called nudification apps — AI-driven applications that generate non-consensual intimate imagery, often used in coordinated harassment campaigns. France has championed an outright ban with criminal sanctions, supported by Germany, Italy, Belgium and Spain. Negotiators are exploring whether the ban can be folded into the omnibus or whether it requires a dedicated legislative track. Either way, MEPs intend to deliver a clear prohibition by the end of 2026.
Other elements of the omnibus
The digital omnibus also addresses: streamlined cybersecurity reporting under the NIS2 Directive, simplified small-business compliance under the Data Act, technical fixes to the Cyber Resilience Act for open-source software, and clarifications on the GDPR-AI Act interaction. The aim is to reduce regulatory friction on European startups and SMEs while preserving the substantive protection logic of the digital rulebook. The Mario Draghi report on European competitiveness is repeatedly cited as the political mandate for these adjustments.
The press conference
If a deal is reached, a press conference with the lead MEPs is foreseen on Wednesday at 11:00. Parliament rapporteur Brando Benifei (S&D, Italy) and shadow rapporteur Dragos Tudorache (Renew, Romania) — the original AI Act co-rapporteurs — are leading negotiations from the Parliament side. Council is represented by the Belgian Presidency in coordination with permanent representatives. The outcome will set the tone for AI governance in the EU through the rest of the legislative cycle.
