In this post Eliezer Yudkowsky argues that the existential risk posed by artificial superintelligence cannot be mitigated by individual action, corporate self-regulation, or localised prohibitions, but only by coordinated international law — specifically, a global treaty restricting the specialised hardware used to train and run frontier AI systems — and he systematically dismantles the notion that extralegal or violent resistance would be effective, on the grounds that shutting down any single company, researcher, or national datacenter does nothing to change the overall trajectory, while simultaneously making the case that lawful, predictable, avoidable state force is a categorically different thing from the chaotic violence some critics conflate it with.

https://x.com/ESYudkowsky/status/2043601524815716866

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit askwhocastsai.substack.com/subscribe