An open letter calling for a six-month pause on giant AI experiments collects 30,000 signatures and changes nothing

On 29 March 2023, fifteen days after the release of GPT-4, the Future of Life Institute published an open letter titled Pause Giant AI Experiments. The letter argued that AI systems with human-competitive intelligence "can pose profound risks to society and humanity" and called for an immediate pause of at least six months on the training of AI systems more powerful than GPT-4.

The signatories list moved fast. Within 48 hours, as reported by The Guardian and Wired, names attached to the letter included Elon Musk, Apple co-founder Steve Wozniak, deep-learning pioneer Yoshua Bengio, AI textbook author Stuart Russell, and Skype co-founder Jaan Tallinn. By the end of the week the count had crossed thirty thousand. None of the chief executives of OpenAI, Anthropic, Google DeepMind or Microsoft signed.

Why the pause did not happen

The operational ask, a coordinated halt across competing labs in different jurisdictions, would have required either binding antitrust-cleared agreement or a regulator with enforcement reach across at least the United States, the European Union and the United Kingdom. Neither existed. As MIT Technology Review pointed out in coverage published a week later, the letter's structural problem was not its diagnosis but its prescription: the closest thing the field had to a coordinating body was a chat group of safety researchers, and they were not the ones running the training clusters.

Within the labs themselves, the response was generally unimpressed. Anthropic's chief executive Dario Amodei, in an interview with Time later that spring, said the letter "didn't really engage with the actual questions about how to make AI safe." Sam Altman, asked at MIT in May, dismissed the framing while accepting parts of the underlying critique. The letter's most concrete legacy was language: "frontier AI", a phrase that had been used inside research circles, broke through to mainstream policy discourse in the weeks after publication.

It did not stop a training run. It put a phrase into Hansard.

Three years on, the letter reads as the high-water mark for a particular kind of public AI advocacy: long-horizon, signature-driven, addressed to nobody in particular. The advocacy that followed was narrower, technically engaged, and aimed at specific policy instruments. The Bletchley declaration, the EU AI Act's general-purpose chapter, and the network of national AI Safety Institutes all carry vocabulary the letter helped popularise. None of them were what the letter asked for.