EU member states and Parliament agree to delay high-risk AI Act enforcement to December 2027 as Europe softens its landmark AI regulation under Big Tech pressure
EU member states and European Parliament lawmakers reached a provisional agreement on 7 May 2026 to weaken and delay elements of the EU AI Act — the world's most comprehensive artificial intelligence regulation. The agreement, which still requires formal endorsement by EU governments and the Parliament, was reached after nine hours of negotiations. The most significant change is a delay to the enforcement of rules governing high-risk AI systems — including those involving biometrics (technology that identifies individuals from physical characteristics), critical infrastructure, and law enforcement — from the original August 2026 deadline to December 2, 2027. The delay is part of the European Commission's broader push to simplify a range of new digital rules and reduce compliance burdens on businesses. Critics have described the agreement as Europe 'caving to Big Tech', arguing that delaying the high-risk provisions removes near-term accountability for some of the most consequential AI applications. Supporters contend that giving businesses more time to adapt will lead to better compliance outcomes. Even with the amendments, the EU AI Act remains the strictest AI regulatory framework in the world. The provisional deal has direct implications for UK firms advising EU clients: the August 2026 compliance window that had been driving urgent advisory mandates on AI governance and high-risk system audits is now extended by 16 months, reshaping the near-term regulatory advisory pipeline.
Why this matters
The delay to EU AI Act high-risk enforcement is material for law firms that have been building out regulatory advisory practices around the August 2026 compliance deadline. The 16-month extension to December 2027 effectively redistributes a concentrated wave of compliance work — AI governance frameworks, conformity assessments, and high-risk system documentation — over a longer period, which may reduce short-term revenue pressure for compliance teams but also extends the advisory opportunity. UK firms advising EU-based clients on AI Act compliance will need to revise their project timelines and re-scope the urgency of current mandates. The 'why now' trigger is explicit: the Commission's drive to simplify digital regulation combined with sustained industry lobbying from large technology companies has created political pressure to ease implementation timelines.
On the Ground
A trainee on an EU AI Act compliance matter would update the regulatory impact assessment memo to reflect the new December 2027 enforcement deadline for high-risk AI systems, revise the client's AI governance policy drafting timeline accordingly, and assist with vendor due diligence questionnaires to assess whether AI tools used by the client fall within the high-risk category as newly delayed.
Interview prep
Soundbite
The 16-month enforcement delay reshapes the entire EU AI compliance advisory pipeline — firms must re-scope live mandates now.
Question you might get
“Under the EU AI Act, what makes an AI system 'high-risk', and what obligations does the Act impose on developers and deployers of such systems — even after the enforcement delay?”
Full answer
EU governments and the European Parliament have agreed to delay enforcement of the AI Act's high-risk AI rules from August 2026 to December 2027, the most significant rollback of the world's strictest AI regulation since it came into force. This matters for law firms because the August 2026 deadline had been a significant driver of regulatory advisory mandates — AI governance frameworks, conformity assessments, and high-risk system audits were all being scoped around that date. The extension gives clients more time but also changes the commerciality of those mandates: urgency fees compress, and multi-year advisory relationships replace sprint-style compliance projects. The structural trend is a tension between regulatory ambition and industry pushback, which suggests further softening of the AI Act's enforcement posture is possible before December 2027. This will be a live issue for every firm advising EU-regulated businesses with AI deployments.
My notes
saved