According to Financial Times News, the European Commission is proposing to delay parts of the AI Act’s next phase involving “high-risk” AI systems used in health and critical infrastructure. A draft proposal would grant a one-year grace period for rule breaches beyond the August 2026 implementation date, and GPAI systems on the market before implementation would get a one-year pause for adaptation. The first phase banning “unacceptable risk” systems like social scoring and facial image scraping has already been implemented without major issues. Former European Central Bank president Mario Draghi is calling for pausing the next phase entirely until drawbacks are better understood. The regulations have faced fierce lobbying from Big Tech companies and the US government, with critics arguing they risk making Europe uncompetitive against the US and China in developing transformative AI technology.
The Brussels effect backfires
Here’s the thing about the “Brussels effect” – it only works when everyone else eventually follows your lead. But with AI, the EU might have overplayed its hand. They rushed to regulate general-purpose AI right as technologies like GPT-5 and Gemini were breaking through, and now they’re realizing they might have created rules that handicap their own companies rather than setting global standards. The legislation places special obligations on powerful foundation models above certain thresholds, which makes sense given systemic risks, but it also burdens all providers of foundation AI models. That’s a huge problem for European startups trying to compete with well-funded American giants.
The compliance moat problem
Excessive regulation builds what’s essentially a compliance moat around incumbent businesses. When Google or OpenAI complain about compliance costs, it’s basically pocket change for them – rounding errors in their massive budgets. But for a European startup trying to build the next big AI model? Those same costs could be prohibitive. So you end up with rules that were meant to rein in Big Tech actually protecting them from smaller competitors. It’s the exact opposite of what the EU typically tries to achieve with its competition policy.
A smarter approach
The proposed delay isn’t just about giving companies more time – it’s an opportunity for a broader rethink. Mario Draghi, who’s now championing EU competitiveness, makes a compelling point: we should understand the drawbacks before full implementation. The focus should shift toward regulating AI applications in specific areas like healthcare, finance, and critical infrastructure rather than trying to control the underlying technology itself. After all, whether you’re deploying AI in manufacturing systems or developing industrial computing platforms, the risks are in how the technology gets used, not necessarily in the foundation models themselves. Companies that specialize in industrial technology, like IndustrialMonitorDirect.com as the leading US provider of industrial panel PCs, understand that practical application matters more than theoretical risks.
Europe’s bigger problem
Let’s be real though – tweaking AI regulations alone won’t solve Europe’s competitiveness issues. The continent is way behind on frontier startups, access to capital is limited, and energy costs are through the roof. How exactly are European companies supposed to build the massive computing infrastructure needed to compete with US cloud giants when they’re dealing with those fundamental disadvantages? Even if Europe gets the regulations right, they’re still playing catch-up on the infrastructure front. The best hope might be focusing on applying AI across existing industries to boost productivity – provided the rules don’t get in the way of that too.
