OpenAI introduces new ‘Trusted Contact’ safeguard for cases of possible self-harm https://t.co/2Ioqdexkib ...