Under its evaluation agreement with OpenAI, Apollo Research deployed a team of six evaluators for a four-month embedded audit of model alignment and control mechanisms

Activity

Activity (0)
Loading comments...
Opening sign-in…