“AI policy researcher Miles Brundage speculated that senior White House officials made the decision against Anthropic without input from domain experts at the Cyber AI Safety Institute (CAISI) or the National Security Agency (NSA).”