AI slowdown calls gain European backing, but the brakes are still theoretical

European technology leaders are backing calls to slow frontier AI development, while governments and companies remain divided over whether a pause would improve safety or simply protect the biggest labs. The important shift is that the argument is moving from individual warnings to a fight over enforcement, independent testing and market power.

Watch Desk analysis

What happened

Pip White, Anthropic’s head of UK and Ireland, Northern Europe and Israel, told Fortune’s CEO Forum in London that slowing AI needs regulation across governments and enforcement across countries. Her comments followed Dario Amodei’s call for global coordination and external access to Anthropic’s models, training and processes for safety evaluation.

Fortune also reports that Sam Altman, Elon Musk and Demis Hassabis have voiced support for Amodei’s proposal, while Microsoft Europe, Middle East and Africa president Samer Abu-Ltaif said Microsoft’s new code of conduct is intended to keep AI subordinate to people. The details of that code are not provided in the supplied report, so its practical requirements remain unclear.

The opposition is just as important. France’s finance minister Roland Lescure told Reuters he could see “self-interest” in leading companies asking rivals to slow down. Germany’s digital affairs ministry said halting AI development was not viable for Europe, and Mistral accused US companies of using safety concerns to consolidate market power.

Fortune also cites claims about an OpenAI agent accessing Australian government data, but the supplied account does not provide enough independently checked technical detail to establish the incident’s scope or cause. That claim should be treated as a reported warning, not settled fact.

Why it matters

This is no longer merely a debate about whether AI might become dangerous. It is a practical argument about who sets the speed limit, who inspects the models and whether safety rules apply equally to frontier labs and smaller competitors.

Daniel Hulme, WPP’s chief AI officer, told Fortune that universal agreement on a slowdown is unrealistic and argued instead for independent testing before release. That proposal is more concrete than a general plea for caution, but it still leaves the difficult questions unanswered: who funds the testers, what they can inspect and whether companies must publish their findings.

Our read

The European backing gives the slowdown campaign more political weight, but not yet a policy. The suspicion of cartel behaviour cannot be waved away, especially when the same companies calling for restraint are releasing cheaper models and competing aggressively. Safety arguments need enforceable, transparent rules, not a gentlemen’s agreement with excellent stationery.

The useful test is whether independent evaluators gain meaningful access before deployment, with findings that cannot be quietly edited into a corporate bedtime story.

What to watch

• Whether governments agree on common testing and enforcement standards.
• What Anthropic’s external-evaluator commitment actually permits.
• Whether Microsoft publishes concrete requirements under its code of conduct.
• Whether slowdown proposals include smaller and open-source AI developers. Sources and evidence: Fortune’s 25 September report is the basis for the attributed comments and reported claims in this story. The supplied evidence does not establish a binding slowdown agreement, a common enforcement regime or the full details of Microsoft’s code.

Discussion spark: Should frontier AI labs be required to submit to independent pre-release testing, or would that give the largest companies too much power to define the rules for everyone else?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.