In the face of an impossible dilemma, the EU appears to have chosen sides. In the name of competitiveness—and under geopolitical threats delivered with the current US administration’s customary subtlety—the bloc has leaked plans to weaken AI guardrails built to preserve its democratic nature.

Brussels is preparing to loosen the corset it only recently buckled around artificial intelligence. A draft “simplification package”, due for adoption on November 19th, would pause and soften parts of the EU’s vast digital rule book, including the AI Act that entered into force last year. Officials say the respite will help Europe match the United States and China in the race for new technology.

The timing is no accident. Big Tech has lobbied hard; Washington has pressed harder. US President Donald Trump warns that heavy-handed rules could bring trade retaliation or even curb intelligence and weapons flows to Ukraine. To ease tensions, Brussels struck a provisional trade truce with the White House in August. Few in the Berlaymont want to test Mr Trump’s temper again.

Although the AI Act became law in August 2024, most of its obligations—especially for “high-risk” systems that threaten health, safety or fundamental rights—take effect only in August 2026. That lag offers room for second thoughts. A senior EU official told Financial Times Brussels has been “engaging” with the Trump administration on tweaks to the Act and other digital statutes.

You might be interested

Adjustment? Abdication?

The draft, the newspaper says, hands firms breaking the toughest rules a one-year “grace period” before penalties bite. Providers of generative AI already on the market before the deadline would enjoy the same reprieve “to adapt their practices without disrupting the market”. The Commission is also floating a delay in fines linked to new transparency duties, shifting enforcement to August 2027.

Officials argue the pause will “provide sufficient time for adaptation of providers and deployers” (evoking Neville Chamberlain’s “peace for our time”) and trim compliance costs. Paperwork would shrink; policing would tilt towards the Commission’s own AI Office. The text now circulates among member-state envoys. Even if adopted, it must win backing from a majority of EU governments and the European Parliament. Changes may yet slip in.

Meta, owner of Facebook and Instagram, says Europe risks walling itself off from cutting-edge services if it refuses to bend. Other firms echo that plea. A Commission spokesperson confirms that “various options are being considered”, but insists Brussels remains “fully behind the AI Act and its objectives”. Writing world-leading rules proved hard; enforcing them may prove harder still.

Democracy in the balance

Into this dispute steps a bracing warning. “Democracy’s survival depends on a shared commitment to truth, voice and consequence,” wrote Eliot Higgins, founder of Bellingcat, and Natalie Martin, a political scientist, in a November report for the Demos think-tank. They see an “epistemic emergency” spawned by social media and digital platforms and urge urgent “democratic repair”.

Democracy, they argue, rests on verification, deliberation and accountability—pillars that cannot stand without trustworthy information. Europe still enjoys stronger foundations than elsewhere, so repair remains possible. Yet the region could squander that edge if economic priorities clash with democratic needs.

(Deferral of the EU AI Act) will provide sufficient time for adaptation of providers and deployers. — unnamed EU officials as quoted by Financial Times

Mario Draghi, a former Italian prime minister, jolted leaders last year by exposing Europe’s yawning productivity gap, rooted in technology. Ever since, officials chant the mantra of “catching up”. But Mr Draghi’s alarm also fuels calls to slim rule books. If deregulation becomes reflex, it may entrench precisely the tools that erode democracy.

Rules, not riddance

Industry now frames regulation as a growth killer. That view has seeped into Brussels’s omnibus deregulation bills and drives the current push to suspend parts of the AI Act. The tech lobby attacks data-privacy rules with similar zeal. Continuous audits of regulation make sense, but policymakers risk forgetting a basic truth: some activities deserve to shrink because they harm the public realm.

Debates on the AI Act often blur the line between overseeing technology itself and restraining its uses. Tuning the first may be wise. Yet no economic case exists for allowing manipulative or exploitative practices. A ban on real-time biometric recognition, for instance, may cripple firms that sell it. That is a feature, not a flaw. Lawmakers should keep debating what counts as harm, but if they judge an activity harmful, its absence is no reason to scuttle the rule.

Democracy’s survival depends on a shared commitment to truth, voice and consequence. — Eliot Higgins, founder of Bellingcat

The spread of large language models, shaped by Silicon Valley’s “tech bro” instincts, poses a similar test. Do models that nudge voters or homework assignments truly raise productivity? If not, Europe must foster home-grown alternatives. Projects such as the digital euro and the EU identity wallet hint at a different path, combining smart rules with public procurement to create demand for democracy-friendly technology.

Trade-offs and trade deals

Another trap lies in treating the digital economy as a bargaining chip. Washington wants softer EU tech rules; Brussels wants tariff relief. The summer truce shows how easily one can be traded for the other. Yet surrender carries a price for democracy.

Those who gain most when Europeans chant “rules bad, innovation good” sit far from the Rhine or the Danube, the report claims. American platforms enjoy natural-monopoly perks: first-mover advantages and the ability to extract rents. That should temper Europe’s eagerness to woo them. Britain’s boast about landing flashy new data centres illustrates the risk, the authors write. The sheds may merely subsidise foreign firms that weaken local economies while polluting Europe’s information sphere.

A wiser course pairs productivity with democracy by nurturing European digital services. Catching up counts, but only if Europe keeps a clear head about what it is trying to catch. The Commission’s looming vote on its simplification package will show whether the bloc can balance growth pressures against the need to protect “truth, voice and consequence”. The exercise may decide not just the future of the AI Act but the health of European self-government itself.