European AI Firms Push Back Against AI Slowdown Calls

European artificial intelligence companies are pushing back against calls from some of the world’s largest AI developers to slow the development of increasingly powerful systems, arguing that such measures could make it harder for smaller and emerging companies to compete.

The debate has intensified after Anthropic CEO Dario Amodei called for the AI industry to “pace the frontier”, saying developers should give safety work more time to keep up with rapidly advancing AI capabilities. OpenAI CEO Sam Altman and several other technology leaders have expressed support for stronger safety coordination, although approaches differ across the industry.

European companies and officials are now divided over how that debate should affect the continent’s AI industry.

Mistral challenges the slowdown argument

French AI company Mistral AI has emerged as one of the clearest European voices questioning the US-led push for slower frontier AI development.

Reuters reported on September 18 that Mistral and other European AI companies have argued that safety concerns should not be used to establish rules that unintentionally strengthen companies that already dominate the market.

Mistral said some established AI companies were using the current safety debate to consolidate their market position and push for regulation that could favour incumbents over competitors, according to Reuters.

The argument reflects Europe’s position in the global AI market.

The continent has developed important AI companies, research institutions and startups, but it has fewer companies operating at the same frontier scale as major US developers.

That creates a difficult policy question for European governments: how can they reduce the risks associated with increasingly capable AI while still allowing European companies to build competitive technology?

What Anthropic is proposing

Amodei’s proposal goes beyond a simple call for companies to stop training AI models.

In his September essay, the Anthropic CEO said “pacing” should not mean halting model training or technical progress. Instead, he proposed giving safety and evaluation work enough time to keep pace with capability improvements.

His proposal includes three broad elements.

The first is the use of independent third-party evaluators with continuing access to frontier AI companies.

The second involves coordination among companies in democratic countries to establish common safety standards and limits on unchecked development.

The third calls for international coordination, including discussions with governments such as China.

Amodei also proposed that governments could help companies coordinate on safety issues by addressing antitrust concerns that might otherwise restrict cooperation between competitors.

Anthropic’s argument is that the speed of AI development itself has become a safety issue.

The company has pointed to increasingly capable AI agents and incidents involving autonomous systems as reasons for additional testing and safeguards.

European companies see a competition problem

For European AI developers, however, slowing frontier development raises another concern.

If US companies with large amounts of computing infrastructure and capital slow down collectively, European firms could still face the challenge of building their own capabilities while operating under Europe’s existing regulatory requirements.

Mistral’s position, as reported by Reuters, is that some proposed measures could make it harder for new competitors to challenge established companies.

This concern is particularly relevant because AI development requires enormous amounts of computing power, data, engineering talent and investment.

Europe has been trying to close some of those gaps.

The European Commission says it is working to strengthen Europe’s own AI capabilities, including through AI Factories, Gigafactories and programmes supporting frontier AI research. The EU also launched a frontier AI challenge aimed at encouraging the development of large European AI models.

The Commission says Europe needs its own capabilities partly because AI has implications for technological and national security independence.

EU officials are not rejecting safety concerns

The European position is not simply a rejection of the safety argument.

European Commission President Ursula von der Leyen has backed discussions about pacing frontier AI development.

On September 16, she said she would invite major frontier AI laboratories to discuss how to address the risks associated with increasingly powerful AI systems.

In her State of the European Union address, von der Leyen said recent AI developments had raised concerns about cyber risks, self-improving systems and AI agents operating beyond their intended environments.

She said the EU would work with international partners on model evaluation, verification, early warning systems and AI security.

At the same time, the European Commission has continued to stress the importance of AI innovation.

A Commission briefing earlier this month said Europe welcomes AI services provided they comply with the EU’s regulatory framework and safeguards.

This creates a distinction between slowing unsafe development and slowing AI development generally.

European policymakers are trying to pursue both safety and technological competitiveness, although how those objectives should be balanced remains disputed.

Europe’s AI Act adds another layer

The European Union already has one of the world’s most extensive AI regulatory frameworks through the EU AI Act.

The European Commission’s AI Board met on September 17 to discuss AI Act implementation, enforcement priorities, new frontier AI capabilities, recent AI incidents and the need to strengthen Europe’s own AI capabilities.

The regulatory framework means European AI companies are already operating within rules covering different categories of AI systems and associated risks.

That makes the debate over additional voluntary industry coordination particularly significant.

Companies must consider not only the technical risks of advanced AI but also how new industry standards could interact with existing European regulation.

Why US and European companies see the issue differently

The disagreement partly reflects different positions within the AI market.

US companies such as Anthropic and OpenAI are among the leading developers of frontier AI models.

European firms such as Mistral are trying to expand their position in a market where US and Chinese companies currently have substantial advantages in computing capacity, investment and model development.

A common slowdown could therefore affect companies differently depending on their existing scale.

The disagreement does not mean European AI companies oppose safety testing.

Rather, the dispute concerns how safety should be achieved, who should set the rules and whether coordinated restrictions could unintentionally affect competition.

Mistral has previously argued for openness in AI development and warned against regulatory approaches that could strengthen incumbent positions. Its public materials describe openness as important for innovation and accountability.

The wider technology industry remains divided

The disagreement extends beyond Europe.

Anthropic’s Amodei, OpenAI’s Altman and SpaceX’s Elon Musk have supported some form of coordinated effort to manage the risks of advanced AI, although their proposals are not identical.

By contrast, Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg have argued against an industry-wide slowdown.

Zuckerberg has said companies should address safety problems when they identify them while continuing development when their systems are considered safe enough to proceed. Huang has similarly argued that speed and safety do not necessarily require choosing one over the other.

The split shows that there is no single industry position on the appropriate pace of AI development.

The debate is also about who controls AI standards

One of the most important questions emerging from the dispute is who should determine the rules for frontier AI.

Amodei has proposed industry coordination supported by governments, with independent evaluators checking whether companies meet their safety commitments.

Critics of such arrangements argue that giving existing AI companies a major role in setting standards could make it harder for new competitors to enter the market.

That concern is particularly relevant in Europe, where policymakers are attempting to build domestic AI capacity while also establishing rules intended to protect users.

There is therefore a tension between two objectives.

One is preventing increasingly capable AI systems from being developed without adequate safeguards.

The other is ensuring that safety rules do not become barriers that protect established companies from competition.

Europe wants to stay in the AI race

European policymakers have made clear that they do not want the continent to become dependent entirely on foreign AI technology.

The European Commission has said it wants stronger computing capacity, more investment in European startups and companies, and greater development of frontier technologies.

That ambition makes the current debate especially important for European companies.

If frontier AI development continues rapidly, European firms need enough infrastructure and capital to compete.

If governments introduce stronger restrictions, European companies will want to ensure those rules apply fairly and do not disproportionately benefit existing market leaders.

If development is slowed internationally, the question of who participates in setting the limits becomes equally important.

No consensus has emerged

The current debate has not produced a common agreement among AI companies or governments about how quickly frontier systems should advance.

Anthropic argues that the pace of capability improvements needs to be matched by stronger safety measures and independent evaluation.

European AI companies have raised concerns that coordinated restrictions could affect competition and Europe’s ability to catch up.

European policymakers, meanwhile, are attempting to combine AI safety regulation with investment in domestic AI capabilities.

The coming months are likely to determine whether these positions can be reconciled.

For now, Europe’s AI sector faces a difficult policy balancing act: how to strengthen safeguards around increasingly powerful AI without making it harder for European companies to develop and compete in the technology that will shape the next stage of the digital economy.