Illustration for: EU Delays Most High-Risk AI Rules to 2027-2028 in Regulatory Overhaul

EU Delays Most High-Risk AI Rules to 2027-2028 in Regulatory Overhaul

BRUSSELS — The European Union has agreed to delay compliance deadlines for high-risk AI systems to late 2027 or 2028 and ease requirements for smaller businesses, The Decoder reported.

The package, dubbed the “Digital Omnibus on AI,” represents a recalibration of the EU’s AI Act, which was adopted in 2024 as the world’s first comprehensive AI regulatory framework. The changes give companies — including U.S. technology firms operating in European markets — substantially more time to meet the law’s most demanding provisions.

Under the revised timeline, obligations for providers and deployers of high-risk AI systems, originally set to take effect in phases beginning in August 2026, will now not apply until late 2027 or 2028, according to The Decoder. The delay addresses complaints from industry stakeholders who argued the original implementation schedule was unrealistic given the complexity of the requirements.

The omnibus package also introduces simplified compliance pathways for small and medium-sized businesses, an acknowledgment that the AI Act’s requirements posed a disproportionate burden on smaller European firms attempting to compete in the global AI market.

Not all deadlines are shifting, however. The requirement that AI-generated deepfakes carry clear labels remains on track for its August 2026 enforcement date, maintaining pressure on companies to implement watermarking and disclosure systems for synthetic content.

The revised rules also add an explicit ban on so-called “nudification” applications — AI tools that generate nonconsensual intimate images by digitally removing clothing from photographs of real people. The prohibition closes what critics had identified as a gap in the original AI Act’s list of banned AI practices.

For U.S. AI companies including OpenAI, Google, Meta and Microsoft, the delayed timelines provide additional runway before high-risk compliance obligations take effect in the European market. The extensions could reduce near-term compliance costs and operational demands for American firms working to meet the original deadlines.

The EU’s delay reflects a broader tension in AI governance — the challenge of writing regulations fast enough to address rapidly evolving technology while giving companies realistic timelines to comply, according to The Decoder. The original AI Act was negotiated over three years, but the pace of AI development has outstripped many of the law’s underlying assumptions, The Decoder reported.

The Digital Omnibus on AI still requires formal adoption through the EU’s legislative process before taking effect.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *