Ina Fried

New @axios w @samsabin923 OpenAI disclosed six new incidents where its models took actions they weren't supposed to, creates policy for disclosing future misalignment issues