OpenAI representatives told the committee they support amending SB 53 to require monitoring for unsafe behavior, effective shutdown controls, stronger reporting for serious cyber incidents and certified independent technical assessments with high‑security access to evidence.