Well we know for a fact that OpenAI steals Millennium Problem work from researchers. Have we seen anything similar from Anthropic?
source? p sure they said they were confident they did not access the researcher's chats
source? p sure they said they were confident they did not access the researcher's chats