AI's Biggest Names Split Over Dario Amodei's Slowdown Call

The News

Dario Amodei, the Anthropic chief executive, has triggered the loudest AI-policy row of the year with a Saturday-morning essay titled "We Must Pace the Frontier", which argues that the breakneck pace of frontier AI development should be deliberately slowed. Within days rivals, politicians and former officials had all lined up to agree or object, according to The Verge's running tally of the reactions.

Amodei set out three steps: embedded third-party evaluators who can verify whether a company is honouring its safety commitments and report incidents; coordination among frontier labs in democratic countries on shared standards and limits; and wider coordination between democratic and authoritarian governments "to the extent this is possible". He was careful to add that pacing "does not mean halting model training", and said Anthropic is "unilaterally committing" to the first step.

The endorsements came fast. Sam Altman of OpenAI said he agrees "that we need to pace the frontier", backed independent evaluators as a "great idea", and welcomed "a federal framework that sets consistent safety requirements". Google DeepMind co-founder Demis Hassabis said the essay "points towards the right path forward". Elon Musk was blunter still, reposting it with the words "Dario is right".

Why It Matters

The striking thing is not the safety rhetoric, which the industry has recycled for years, but the near-unanimity among commercial rivals who normally agree on almost nothing. When the people who profit most from a technology ask to be reined in, scepticism is healthy, and critics supplied it. Former Meta scientist Yann LeCun noted that Amodei called an early model too dangerous to release back in 2019, and said the warnings deserve mockery now.

The politics split cleanly. President Donald Trump rejected any slowdown outright, framing the safety push as a conspiracy that only benefits China. Vice President JD Vance called firms begging to be regulated "a bit of a trojan horse", while Bernie Sanders, Kamala Harris and Pete Buttigieg backed a pause. The last time the sector debated slowing down this openly was the 2023 "pause" open letter that gathered thousands of signatures and changed nothing; the difference now is that the lab bosses themselves are driving it.

Indian Angle

For India this is a debate happening one rung above where its own builders currently stand. "Pacing the frontier" is a conversation among labs that already occupy it; India is the ambitious fast follower still climbing. Under the IndiaAI Mission, approved by the Union Cabinet in 2024 with an outlay of ₹10,371.92 crore, the government has subsidised GPU access and tasked startups such as Sarvam AI with building a sovereign foundation model, while Ola's Krutrim became the country's first AI unicorn. A coordinated Western slowdown could hand these firms breathing space, or, if "coordination" hardens into tighter compute and export controls, lock them further out.

There is a regulatory signal too. New Delhi has so far favoured a light-touch, pro-innovation posture, from MeitY's evolving advisories to the still-unnotified rules under the Digital Personal Data Protection Act. With India due to host a major global AI summit in early 2026, it must decide whether embedded evaluators and shared frontier limits are principles it champions or constraints it resists. The instinct of IT minister Ashwini Vaishnaw has been that regulation must not smother a young ecosystem.

FAQ

What exactly is Amodei proposing?

Three measures: embedded independent evaluators to check safety compliance, coordination among frontier labs in democratic countries, and eventual coordination with authoritarian governments. He stresses this means slower progress, not stopping model training altogether, and says Anthropic is adopting the evaluator step on its own.

Does this create a new law?

No. These are voluntary commitments and calls for future frameworks, not binding rules. Altman has asked for a federal safety framework, but former FTC chair Lina Khan noted that existing consumer-protection laws already apply to AI products, so no special exemption exists.

How does it affect Indian AI startups?

Indirectly. A Western slowdown could narrow the capability gap that firms like Sarvam and Krutrim are racing to close, but it could also tighten access to the compute and export pathways they depend on. Much rests on how "coordination" is defined in practice.

Where can I read the original coverage?

The Verge compiled the full set of executive and political statements, with direct quotes from Altman, Hassabis, Musk and Trump, in the report linked below.

This story was reported by The Verge. Read the full original coverage at The Verge.