Apollo Research says it has launched an embedded evaluation pilot with OpenAI, with six evaluators on site for four months
AI summary
Apollo Research announced a pilot with OpenAI for embedded evaluation: six evaluators from Apollo Research will work inside OpenAI for four months, investigating whether models will try to weaken oversight and control over themselves. The pilot starts today.
Full text · Original
Oh and we forgot to mention this in the thread:
**The pilot is starting TODAY**
We're launching a pilot towards embedded evaluations with @OpenAI.
Six evaluators from Apollo Research will work within OpenAI for four months, investigating whether models attempt to undermine their own oversight and control.View the quoted post on X