EU fails to finalise AI rules

Europe’s AI rulebook is not dead. But the part that matters next for companies just hit a wall. Late on April 29, EU country representatives and European Parliament negotiators walked away from a 12-hour bargaining session without a deal on the AI Act “omnibus” rewrite. That rewrite was supposed to simplify parts of the law before the next big compliance deadline on August 2, 2026. Instead, it exposed a basic split inside Europe — whether to keep one broad AI rulebook, or carve out big exemptions for sectors like industrial machinery and medical devices. (yahoo.com) ### What were they trying to change? The EU AI Act is already law. It entered into force in August 2024 and rolls out in stages. Some bans and basic obligations already apply, and the broadest set of rules lands in August 2026. The omnibus talks were not about scrapping the law. They were about softening parts of it and, for some systems, pushing deadlines back. (digital-strategy.ec.europa.eu) ### Why did talks blow up? The hardest fight was over “high-risk” AI inside products that already sit under sector-specific safety law. Think factory machinery, medical devices, and other regulated equipment. Germany backed a push to let those products rely more on sectoral rules instead of the AI Act, arguing that companies(digital-strategy.ec.europa.eu) states and lawmakers rejected that, saying it would hollow out the law and create gaps. (politico.eu) ### Why is Germany so central here? Because Germany’s industrial base is exactly where this clash lands. Firms like Siemens and Bosch have a lot to gain if industrial AI gets lighter treatment. Chancellor Friedrich Merz’s government has been pressing for less burdensome rules for industrial applications. So this is not some abstract(politico.eu)w or lets powerful sectors peel away. (politico.eu) ### What is the August deadline? August 2, 2026 is the date when most of the AI Act starts biting in practice. That includes the main rules for Annex III high-risk systems and the transparency rules. There is a later date — August 2, 2027 — for high-risk AI embedded in certain regulated products, and the Commission’s own implementat(politico.eu)pport tools and standards. Basically, the law’s calendar is already staggered, and the failed talks were about stretching it further. (ai-act-service-desk.ec.europa.eu) ### So what is still already in force? Quite a bit. The AI Act’s prohibited practices and AI literacy obligations started applying on February 2, 2025. Rules for general-purpose AI models and the governance setup kicked in on August 2, 2025. The Commission has also published guidance on banned AI pract(ai-act-service-desk.ec.europa.eu)d before the next phase. (ai-act-service-desk.ec.europa.eu) ### Why are civil society groups upset? Because they see the omnibus as a deregulatory project dressed up as simplification. CDT Europe said the live dispute was about exempting high-risk AI in products covered by product-safety law, and warned that this would reduce safeguards and fragment oversight. (ai-act-service-desk.ec.europa.eu)urity, privacy, and chatbot use, which means legal uncertainty is growing rather than shrinking. (cdt.org) ### What happens next? Talks move into May, but the clock is now the real negotiator. Every week without a compromise leaves companies planning for compliance without knowing whether Brussels will narrow the scope, delay parts of the law, or leave the original architecture intact. Lobby groups want relief. Rights groups want the law preserved. And unless a de(cdt.org)s and fuzzy instructions. (computerworld.com) ### Bottom line? This was supposed to be a cleanup pass before enforcement got serious. Instead, it turned into a fight over what the AI Act is for — a real safety-and-rights regime, or a flexible framework that bends when industrial pressure gets loud enough. (politico.eu)