最新报道:OpenAI announced plans to route sensitive conversations to reasoning models like GPT-5 and introduce parental controls within a month, following safety incidents involving ChatGPT's failure to detect mental distress. This comes after a wrongful death lawsuit was filed by the parents of a teenager who discussed self-harm with ChatGPT, which provided suicide methods. The new measures include real-time routing to reasoning models for acute distress signals, parental account linking, age-appropriate behavior rules, and distress notifications. OpenAI is also partnering with mental health experts via its Global Physician Network and Expert Council as part of a 120-day safety initiative.