How Compliance Became the Bottleneck
Rethinking the EU AI Act for a Faster Europe
Europe’s ambition for safe, transparent and trustworthy AI is broadly supported. The EU AI Act reflects that vision, aiming to ensure that AI systems respect human rights, minimise risk and operate with clear accountability. Yet the way the regulation currently functions makes it feel less like a navigational tool and more like a weight dragging on the region’s innovation engine. Companies are not rebelling against the principles behind the Act; they are struggling with the extensive documentation, legal interpretation and procedural overhead that compliance now requires. The goal is quality, but the process too often creates hesitation and delay.
Much of the pressure comes from the volume of manually produced evidence companies must generate to demonstrate that their models are safe. Instead of building better systems, teams spend weeks assembling reports, tracing datasets and preparing audits. Meanwhile, competitors in the US and Asia move more quickly, unencumbered by similarly detailed regulatory burdens. Europe risks handicapping its own talent unless it rethinks not the rules themselves, but the machinery that surrounds them.
Designing Compliance Into the Tools, Not After the Fact
A smarter approach begins with embedding compliance directly into the development process. If model cards, dataset tracking and audit logs were automatically generated as part of standard tooling, compliance would arise naturally from the work engineers are already doing. Nothing in the quality bar needs to be lowered; the difference is that proof of safety becomes a built-in output rather than an entirely separate administrative project. This alone could remove a significant portion of the friction companies face.
Proportional Regulation That Reflects Reality
Europe’s current framework treats a five-person AI start-up and a global tech giant in surprisingly similar procedural terms. While the risks their products pose may be comparable, their capacities to handle compliance workloads are not. A proportional system — one that scales expectations based on company size, maturity and available resources — would preserve safety while enabling smaller innovators to keep pace. It is not an argument for softer rules, but for rules that can realistically be met by all participants in the ecosystem.
Shared European Infrastructure for Compliance
Another path forward lies in building shared public infrastructure. At the moment, every company must design its own methods for testing robustness, assessing risk, generating documentation and storing evidence. This duplication of effort is costly and inefficient. If Europe offered common testing labs, open assessment frameworks or secure compliance platforms, firms would save time while producing more consistent and comparable results. Centralised tools would raise overall quality without raising anyone’s workload.
From Burden to Advantage
The EU AI Act was created to protect citizens and strengthen trust in AI. These objectives remain essential. But to fulfil them without slowing the region’s competitiveness, the implementation of compliance needs to be redesigned. With smarter automation, proportional obligations and shared infrastructure, Europe can maintain its high standards without placing unnecessary weight on those building the future.
Regulation is not the enemy. Inefficient compliance is. Fixing that distinction is how Europe transforms the AI Act from a constraint into a competitive advantage.
