Source: The Atlantic
Major AI labs are hiring top researchers away from universities with agreements that restrict publication and public scrutiny, effectively privatizing work that was previously peer-reviewed and openly debated. This creates a structural problem: the researchers best positioned to audit AI safety and performance are now contractually prevented from doing so, while companies control what gets published about their own systems. The shift also disadvantages academic institutions that can't compete on salary, concentrating both talent and knowledge toward a handful of private players.