The Pulse
Europe’s AI Firms Reject U.S. Calls to Slow Frontier Development
Mistral and other European AI companies accuse U.S. rivals of using safety proposals to protect their market positions. The dispute comes as Europe tries to close its gap with American and Chinese frontier-model developers.
AI.info Team ·
“Some incumbents are using this moment to consolidate their market position, pushing for regulation designed to favour them over competitors.”
Mistral, in a statement quoted by Reuters
European AI companies and officials are pushing back against calls from leading U.S. developers to slow the development of the most capable models, arguing that safety rules could reinforce the position of companies already ahead. The dispute sharpens a competitive problem for Europe: the region has produced few frontier-model developers and still relies heavily on systems from OpenAI, Anthropic and Microsoft.
Reuters reported on September 18 that French AI company Mistral, Germany’s Black Forest Labs, Swiss privacy company Proton and French-American software company Hugging Face all questioned the case for a broad slowdown. Their objections do not amount to a rejection of safety testing. Instead, they focus on who writes the rules, which companies can afford to comply and whether proposed controls would make it harder for new entrants to compete.
Anthropic’s safety proposal meets European resistance
Anthropic chief executive Dario Amodei has urged advanced AI developers to slow the pace at which they improve model capabilities. In an online essay, he called for leading companies to coordinate on safety standards under an antitrust waiver and proposed permanent embedded evaluators to monitor whether developers comply.
Amodei argues that AI development is moving faster than governments can regulate. He said companies should work together voluntarily on standards while lawmakers pursue formal rules, warning that increasingly capable systems could create severe risks, including threats to humanity.
OpenAI chief Sam Altman and xAI founder Elon Musk have also warned about the dangers posed by powerful AI systems and agreed that development should be paced. European companies and policymakers have taken a more sceptical view of the proposed response, particularly when safeguards would be designed or administered by the largest developers themselves.
Mistral sees a competition problem
Mistral is widely viewed as Europe’s strongest prospect for building a frontier AI model capable of challenging U.S. and Chinese developers. The three-year-old company has a direct interest in continued access to the frontier, but its concern reflects a wider dependence problem: European organizations currently obtain many advanced models from American providers.
Raphael Auphan, chief operating officer of Proton, described the U.S. calls as “totally self-serving” at a technology event in Paris. “They just want to preserve a dimension of dependency on their service,” he said, according to Reuters.
Ben Brooks, head of public policy at Black Forest Labs, warned that creating a formal separation between large laboratories and newer entrants through “arbitrary thresholds will chill open innovation at or near the frontier”. The concern is that compliance requirements tied to model size, computing resources or capability could impose costs that established companies can absorb more easily than startups.
Hugging Face chief executive Clement Delangue made a similar distinction. “It’s not time to slow down but to accelerate,” he wrote on X the previous week, while supporting Amodei’s proposal to place independent evaluators inside AI companies.
Europe’s sovereignty argument
The debate is tied to Europe’s effort to reduce dependence on foreign technology. Germany’s digital affairs ministry said stopping AI development was unrealistic given the need for digital sovereignty. Daniel Abbou, head of Germany’s AI industry association, said European companies should concentrate on closing the gap rather than debating whether the leading developers should pause.
“For European firms, this debate is irrelevant. We first need to catch up on our technological backlog,” Abbou said.
French finance minister Roland Lescure was more direct in his criticism of the American position. “Making everyone behind them slow down so they can stay in first place – I can clearly see their self-interest,” he told Reuters.
European Commission President Ursula von der Leyen said on September 16 that the European Union’s AI rules put the bloc in a position to help shape global efforts to address the technology’s risks. Britain’s King Charles hosted technology and AI leaders at a summit in Scotland on September 17.
The argument is over who controls the safeguards
European resistance does not dismiss the underlying safety concerns. Researchers and major developers have warned for years that more capable systems could cause serious harm without effective controls. The disagreement is over whether the companies building the most powerful models should receive a central role in defining dangerous capabilities, selecting evaluators and setting the pace for competitors.
Kristian Kersting, a professor of AI at Germany’s Technical University of Darmstadt, called Amodei’s proposal “regulatory capture”. He said that if large companies decide what counts as dangerous, choose the inspectors and seek an exemption from competition law, the result would not be genuine regulation.
That tension gives Europe’s position a practical edge. A startup trying to catch up may support independent testing but oppose requirements that make frontier development available only to companies with the largest budgets and computing infrastructure. For European firms, the immediate issue is not simply whether advanced AI needs safeguards. It is whether those safeguards leave room for another company to reach the frontier.