EU tells AI firms to prove safety amid industry slowdown calls

The European Union on Monday told artificial intelligence companies they must demonstrate the safety of their services before operating in the bloc, as debate over the pace of AI development intensified following calls from top industry executives to slow down.

EU Commission spokesperson Thomas Regnier said the bloc supports innovation in AI but will not "compromise on our security standards," according to Anadolu Agency. Companies must prove their services are safe for EU citizens, Regnier said, reaffirming the bloc's stance as key provisions of the EU AI Act take effect.jpost+1

A Weekend of AI Safety Alarms

The EU's statement came days after Anthropic CEO Dario Amodei published a 3,800-word essay on Saturday titled "We Must Pace the Frontier," warning that without a deliberate slowdown, AI could within months be capable of actions that pose severe risks, including leading swarms of agents that could compromise critical internet infrastructure. Amodei called for independent auditors to verify safety standards across major AI labs and urged companies to grant external evaluators "ongoing, employee-like access" to monitor their work.techxplore+2

The call drew swift support from other leading executives. OpenAI CEO Sam Altman agreed on the need to slow frontier-model advances, while Elon Musk endorsed Amodei's position in a post on X, and Google DeepMind Alphabet Inc. ↗ 0.78% chair Demis Hassabis also expressed agreement. The chorus of concern sent AI-related stocks lower ahead of Monday's opening bell, with chip and memory companies particularly affected.nytimes+2

Europe Charts Its Own AI Path

Separately on Monday, a coalition of economists and former policymakers — including Nobel laureate Philippe Aghion and former EU Commissioner Margrethe Vestager — published a Transformative AI Strategy warning that Europe risks "permanent marginalisation" without immediate action.ieu-monitoring

The strategy calls on EU member states to target a 15 percent share of global AI data centre capacity by 2030, up from roughly 5 percent today, requiring an estimated €1.3 trillion in investment and an additional 45 gigawatts of power. Clemens Fuest, president of the ifo Institute and a co-author, noted that European AI model developers "together generate less than 2% of the revenue of just two US competitors".ieu-monitoring

Regulation Meets Urgency

The EU AI Act's high-risk system obligations became enforceable on August 2, 2026, requiring providers to complete conformity assessments, register systems in an EU database, and activate post-market monitoring. Companies face fines of up to €35 million or 7 percent of global annual turnover for non-compliance.cloudsecurityalliance+1

Vestager, a co-author of the new strategy, framed the moment starkly: "We must not allow Europe to become a playing field for foreign powers".ieu-monitoring