EU legislators on Thursday morning agreed to postpone restrictions on high-risk uses of artificial intelligence by more than a year, simultaneously exempting industrial AI applications from the scope of the law. This decision, reached after negotiations that concluded around 4:30 a.m. on Thursday, directly benefits major corporations by reducing regulatory burdens and protecting profit margins.
The agreement was confirmed by the Cypriot presidency of the Council of the EU and the European Parliament. It received support from European Parliament lawmakers and EU countries following what was described as "heavy pressure from industry and capitals."
Capital's Victory
The change represents a significant victory for German capital, with top officials including Chancellor Friedrich Merz advocating for the rollback. Their stated aim was to maintain the competitiveness of tech heavyweights such as Siemens and Bosch within the global market. EU countries supported Germany's demand to prevent a "double regulatory burden" for companies utilizing industrial AI, effectively allowing these corporations to comply only with separate machinery rules rather than the broader AI requirements.
While other industries, including medical devices, were not exempted and remain subject to the AI law, the broad carve-out for industrial applications signals a clear prioritization of corporate interests over public safeguards. The deal marks the first significant rollback of rules in the digital space, occurring as the EU faces pressure from the U.S. regarding its tech laws and warnings from its own industry and governments about a perceived disadvantage in the global AI race.
Commission President Ursula von der Leyen stated the deal "provides a simple, innovation-friendly environment" for AI in Europe, while simultaneously claiming to strengthen protections for citizens. This rhetoric frames deregulation as beneficial, despite the actual reduction in oversight for high-risk AI systems.
The State's Complicity
Restrictions on high-risk AI, originally set to commence in August of the same year the deal was struck, will now be delayed until December 2027, extending the period during which these systems can operate with reduced oversight. The EU AI Act itself became law in August 2024, now in its second year, with its phased rollout already anticipating rules governing high-risk uses to kick in this August. The postponement effectively grants capital an additional year and four months of unregulated operation for these systems.
Companies will also receive a grace period of three months, rather than the six months initially proposed, for meeting new requirements to watermark AI-generated content. This minor concession stands in contrast to the extensive exemptions granted to industrial AI.
Empty Promises of Protection
Amidst the broad deregulation, the deal does include a ban on AI systems capable of generating sexualized deepfakes of "identifiable" people, following global outrage over the abusive use of Elon Musk’s AI tool Grok. AI systems that generate child pornography will also be banned. These specific prohibitions, while presented as protections, address only the most egregious and publicly visible abuses, leaving the broader structural issues of high-risk AI deployment in industrial settings largely unaddressed and unregulated, thereby preserving the conditions for capital accumulation at the expense of comprehensive public safety and worker protections.