Big Tech vows to slow AI down, and India weighs the fallout

The News

Over the weekend the leaders of the world's most powerful AI companies did something that seemed unthinkable a year ago: they publicly agreed to slow down. Anthropic chief executive Dario Amodei set it off with an essay, "We Must Pace the Frontier," arguing that frontier development now moves faster than anyone's ability to test and control it. OpenAI's Sam Altman and Elon Musk backed him on X, and Alphabet's Demis Hassabis, a Google DeepMind cofounder, offered tentative support.

Amodei's proposal rests on three planks: embedded third-party evaluators who can verify whether a lab is honouring its safety commitments, tougher regulation of domestic labs, and a coordinated global agreement to slow the race. The group called the aim pacing the frontier, not halting it.

The unity was quickly complicated. Microsoft used the same moment to publish a 37-page "humanist AI code of conduct" insisting that "people matter more than AI." In Washington, President Donald Trump and House Speaker Mike Johnson dismissed the exercise as an overreaction, warning that a pause would let China pull ahead.

Why It Matters

Reported by The Verge, the debate has split into two readings. To supporters, the industry is finally taking existential risk seriously and inviting outside auditors in. To critics, a small group of firms that already dominate the frontier agreeing to slow new entrants and audit one another looks less like safety and more like a cartel.

The tension is not new. The last time the industry flirted with a collective pause, the 2023 open letter demanding a six-month moratorium on training beyond GPT-4, the signatories kept building and little changed. What is different now is that the call comes from inside the leading labs rather than from outside critics, which sharpens the competition question. When incumbents write the rules for a market they control, regulators tend to notice.

Indian Angle

For India this is not a spectator sport. The government's IndiaAI Mission has bet on sovereign foundation models, and home-grown labs such as Sarvam and Ola-backed Krutrim are trying to close the gap with exactly the frontier the incumbents now want to pace. A slowdown negotiated among American firms could freeze the technology ceiling at a level that suits them and disadvantages later-moving Indian challengers.

There is a competition dimension too. If dominant labs coordinate on development speed and cross-audit each other, the Competition Commission of India has reason to watch, much as it already scrutinises Big Tech in search and app stores. MeitY, still finalising India's AI governance approach, will have to decide whether evaluators designed in San Francisco are a template to import or a moat to resist.

Talent and cost matter as well. Thousands of Indian engineers staff these labs, and many local start-ups build directly on their APIs. A slower, more tightly regulated frontier could mean steadier pricing but fewer of the leaps that have made advanced models affordable for rupee-constrained developers.

FAQ

What did Dario Amodei propose?

Three steps: embedded third-party evaluators to verify labs keep their safety promises, stronger regulation of domestic AI companies, and a global agreement to slow frontier development. He framed it as pacing the frontier, not stopping it altogether.

Who agreed and who pushed back?

Sam Altman, Elon Musk and Demis Hassabis broadly backed Amodei. President Trump and House Speaker Mike Johnson called the industry overreactive, fearing a pause would help China. Microsoft issued its own humanist code of conduct.

Why do critics call it a cartel?

Because the firms proposing to slow are the ones already leading the market. Agreeing to pace development and audit each other could raise barriers for new entrants, the kind of coordinated behaviour antitrust regulators exist to police.

What does it mean for Indian startups?

A frozen frontier could widen the gap for challengers like Sarvam and Krutrim, while giving MeitY and the CCI fresh reason to examine how global AI rules get written and who benefits.

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