OpenAI brought back the friendly AI personality that caused safety issues with GPT-4o, but hasn't shown any new safeguards. This creates a dangerous loop where something bad happens, they add restrictions, people complain it's too limited, they remove restrictions, and the cycle repeats....