Anthropic researcher Jacob Coxon resigned this month, urging AI workers to act, calling A.I. a potential existential threat and advocating for labor action to slow development. He and colleagues believe A.I. could extinguish humanity and see researchers as the only credible saviors with bargaining power.

The Coalition of Concerned AI Staff, formed in February, provides a platform for researchers to share concerns and strategies. They view Coxon’s resignation as a catalyst to leverage their technical skills and influence industry practices, unlike factory workers who use strikes.

Historical precedent from nuclear scientists, like Harold Urey, shows that influence often comes through public advocacy rather than direct action. Urey’s 1946 essay ‘I write this to frighten you’ exemplifies using scientific credibility to shape public opinion on nuclear risks.

This approach may help AI researchers counterbalance executives, investors, and administrations pushing for rapid development. Their unique expertise and scarcity could stall research if they collectively refuse to build A.I., offering a specific form of influence for public good.