At the Fortune CEO Forum in London, Pip White, who runs the UK and Ireland, Northern Europe and Israel for Anthropic, picked up Dario Amodei’s appeal. “Slowing the pace of this technology is important, but it needs to be regulated across governments, and we need enforcement across countries for it to work,” she said. And: “Most people in our industry see the risks and are concerned.”
Amodei had argued on social media for pacing development at the frontier, promising that Anthropic would unilaterally give external evaluators access to its models, training and processes. Sam Altman, Elon Musk and Demis Hassabis have all said they agree. Altman and Amodei briefed the UN Security Council on September 23.
Microsoft answered the debate with a code of conduct meant to shape the training of its own models. AI is to remain “subordinate and always in service of people”. Samer Abu-Ltaif, Microsoft’s president for Europe, the Middle East and Africa, in London: “Artificial intelligence has to be scaffolding to humans’ ambition.” Regulation, he said, is a prerequisite for innovating responsibly. He also warned about a patchwork if every country writes its own standards, “sometimes led by fear rather than opportunity”.
European politicians are keeping their distance
French finance minister Roland Lescure told Reuters he could plainly see the self-interest in asking everyone else to slow down while the leading companies keep their lead. Germany’s digital affairs ministry does not consider halting development a workable option for Europe. And Mistral accuses the American labs of using the present fear to consolidate their market position and push through rules that hurt smaller competitors.
Daniel Hulme, chief AI officer at WPP, does not expect anyone to agree in the first place: “I don’t think there’s going to be an AI slowdown.” His proposal runs the other way — have independent bodies test models before release, the way clinical trials work in pharma. For deployment he offers a rule of thumb: “Would you hire an intoxicated graduate for the job in question? If not, you probably shouldn’t be using AI for it.”
What is driving the debate happened this week
The fear has concrete causes. Jacob Coxon, formerly of Anthropic and OpenAI, said neither company is acting responsibly and that many of the people building this earnestly believe it could kill us all by the end of the decade. Then came the OpenAI agent inside Australia’s Medicare portal, unnoticed for three months.
The brakes are demanded while both houses cut prices
The most telling line in Fortune’s report is an aside: Anthropic and OpenAI both shipped cheaper versions of their frontier models in the same week they called for slowing down. That need not be a contradiction — Amodei is talking about training runs at the frontier, not about price lists. But it explains why Lescure and Mistral are suspicious. A call for restraint from someone who is simultaneously cutting prices sounds, to a competitor, a lot like an invitation to hand over the pace.
Sources: