OpenAI says ChatGPT can better detect signs of self-harm and violence as the company faces lawsuits and investigations over dangerous chatbot interactions.
