“However, third-party evaluators may only be one piece of the puzzle of a much larger framework necessary for keeping models in line. [CDT’s] Miranda Bogen, [], told The Deep View….”
thedeepview.com
What AI labs' safety pledges still don't solve
As discussions of an AI slowdown escalate, leaders of AI's top labs may be aligned on where to start: third-party accountability. Leaders from Anthropic, Googl...