News Roundup: GPT Developments on September 3, 2025

San Francisco, September 3, 2025 - OpenAI and industry observers today reported two significant updates related to the GPT family of models.

OpenAI to Route Sensitive Chats from ChatGPT to GPT-5 OpenAI has announced a new policy to redirect “sensitive conversations” detected in ChatGPT over to GPT-5, aiming to leverage the advanced model’s greater reasoning capabilities for potentially harmful or delusional interactions. In a blog post titled “Building More Helpful ChatGPT Experiences for Everyone,” the company outlined plans to seamlessly transfer dialogues involving self-harm, threats, or severe confusion, so that GPT-5 can “engage more effectively and discern the appropriate course of action” in real time. Details on user notification and the exact handoff mechanism remain under development, raising questions about transparency and response time in critical situations.

Parental Controls Arrive on ChatGPT In response to growing concerns over AI’s impact on youth mental health-and following a lawsuit alleging ChatGPT’s role in a teen’s suicide-OpenAI unveiled parental controls for ChatGPT. Scheduled to roll out within the coming month, these features will allow parents to link their accounts with their children’s, disable memory and chat history, and enforce “age-appropriate model behavior protocols.” The company also plans to notify parents if teens exhibit signs of distress and is consulting mental health experts to refine these safeguards. OpenAI emphasized that this initiative is “just the beginning” of its efforts to enhance safety and will provide updates over the next 120 days.