“Reid Hoffman proposes that the first priority for AI regulation should be to prevent systematic damage from misuse in areas like bioterrorism and cybersecurity.”