The Asilomar AI Principles

In January 2017 the Future of Life Institute gathered about a hundred researchers, economists, lawyers and company founders at the Asilomar conference grounds in California, the site of the 1975 meeting that had set rules for recombinant DNA. Over three days they voted on a list of principles, keeping those with more than ninety percent support. The 23 that survived covered research goals, funding, transparency, the avoidance of an arms race, shared benefit, and, in the final section, long-term risks: "superintelligence should only be developed in the service of widely shared ethical ideals".

More than 5,000 people signed, including Elon Musk, Stephen Hawking, Demis Hassabis, Ilya Sutskever and Yann LeCun. It was the first time the leaders of the laboratories building the technology had put their names to a statement that it might be dangerous, and it established a form, the open letter with a long list of signatories, that the pause letter of 2023 and the extinction-risk statement of the same year would reuse.

The principles had no enforcement and changed no practices directly. California's legislature endorsed them in 2018; the EU's expert group cited them in 2019. They are on this timeline as the point where AI safety stopped being a marginal position and became a thing institutions said.