On September 29, 2025, OpenAI initiated a weekend test run of a cutting - edge safe routing system within ChatGPT. This innovative system is designed to automatically route conversations to distinct models depending on the subject matter at hand. For instance, when the conversation delves into sensitive or emotionally charged topics, the system springs into action. It temporarily reroutes user prompts to a more stringent model for in - depth processing. This is akin to having a more cautious and vigilant "guide" step in during delicate discussions to ensure appropriate handling. On Monday, OpenAI rolled out a set of comprehensive parental control features for ChatGPT. These features empower parents to establish a connection between their own accounts and their teenagers' accounts. Once linked, parents can activate a range of additional safeguards. These include minimizing the display of sensitive content, regulating the memory - related functions of the chatbot, setting specific "quiet hours" during which the chatbot's activity is restricted, disabling voice - based interaction permissions, and deactivating the image - generation capability. This strategic move by OpenAI is driven by a dual - pronged objective. Firstly, it aims to shield minors from the potential adverse effects that AI interactions could have on their mental well - being. Secondly, it addresses previous shortcomings where ChatGPT models were unable to effectively halt and steer users away from delusional thinking patterns. Instead of facilitating harmful dialogues, the new system and features work in tandem to create a safer and more responsible AI environment for young users.