On Wednesday, European Parliament lawmakers from the Internal Market and Civil Liberties committees backed a proposal to delay part of the EU’s AI rules on high-risk systems, a move opposed by digital rights advocates and welcomed by the industry. They also used the vote to push for a ban on so-called “nudifier” apps.
MEPs agreed to delay the application of obligations for high-risk AI systems. This includes those used in biometrics, critical infrastructure, education, employment, law enforcement and border management. Instead of August 2026, the rules would now apply from December 2027.
For AI systems falling under existing EU product safety legislation, such as medical devices and radio equipment, the deadline would be pushed further to August 2028. Last week, industry groups warned that they could otherwise face overlapping obligations under both the AI Act and sector-specific rules.
“Companies now need clarity on whether they are high risk or not.” — MEP Arba Kokalari (EPP/SWE)
Lawmakers also supported giving companies more time to comply with transparency obligations for AI-generated content, such as watermarking. However, they opted for a shorter extension than proposed by the European Commission, setting a new deadline of November 2026.
MEP Arba Kokalari (EPP/SWE) stressed the need for clearer and more predictable rules. “Companies now need clarity on whether they are high risk or not”. She added that Europe must make it easier to deploy AI if it wants to remain competitive.
You might be interested
Ban “nudifier” tools
Moreover, as part of the report, MEPs proposed banning so-called “nudifier” applications. These systems use artificial intelligence to generate or manipulate sexually explicit images resembling real individuals without their consent.
The ban would not apply to systems that include effective safeguards preventing such misuse. MEP Michael McNamara (Renew, IRE), co-rapporteur from the LIBE committee, welcomed the compromise as a response to public concerns. “I believe it is something that our citizens expect of the co-legislator”, he added.
More flexibility for businesses
The committees also introduced measures aimed at reducing the regulatory burden on companies. Providers would be allowed to process personal data to detect and correct bias in AI systems, but only when strictly necessary and with appropriate safeguards in place.
Industry voices have also called for swift action. Boniface de Champris, AI Policy Lead at CCIA Europe, urged lawmakers to deliver on promises of simplification. He argues that a pragmatic grace period for generative AI labelling requirements is essential to support innovation. “Adopting a pragmatic 12-month grace period for generative-AI marking and labelling requirements is crucial to show that the EU values innovation over red tape”.
However, civil society groups remain cautious. In a statement published last Tuesday, EDRi warned that while some of the most controversial elements of the Digital Omnibus are being reconsidered, “important risks remain”. The group added that simplification “should strengthen rights, not dilute them”.
Next steps
The European Commission proposed the so-called AI omnibus in November last year, aimed at simplifying the bloc’s rules and improving competitiveness. Following its adoption by the committees, Parliament is expected to vote on its mandate in plenary on 26 March. If approved, negotiations with EU member states in the Council can begin.
