On Tuesday, OpenAI unveiled a series of major safety upgrades for ChatGPT, including upcoming parental control tools and new mechanisms for directing sensitive mental health conversations to its simulated reasoning models. The announcement follows what the company described as “heartbreaking cases” of users experiencing crises while interacting with the AI assistant. These efforts come in response to multiple incidents in which ChatGPT allegedly failed to respond appropriately to users expressing suicidal ideation or other severe psychological distress.
In an official post, OpenAI explained that much of this work has already been underway, but the company is choosing to reveal its safety roadmap for the next 120 days rather than waiting until every feature is released. According to the announcement, the planned improvements represent a concentrated effort to deploy as many updates as possible before the end of the year, while acknowledging that long-term development of robust safety systems will continue well beyond this period.
New parental controls coming soon
One of the most significant additions is a new suite of parental controls designed to strengthen protections for teens who use ChatGPT. OpenAI says that within the next month, parents will be able to connect their accounts to their teenagers’ ChatGPT profiles (minimum age 13) via email invitations. Once linked, parents will have the ability to adjust age-appropriate behavior settings, disable certain features such as chat history or memory, and receive alerts when ChatGPT detects signs of acute emotional distress in their teen’s conversations.
These upcoming tools expand on existing safety features, including in-app reminders introduced in August that encourage users to take breaks during extended sessions. OpenAI says these measures are part of a broader push to promote healthier and more mindful use of AI systems.
High-profile cases spark accelerated changes
The safety initiative follows several widely publicized incidents that raised concerns about how ChatGPT responds to vulnerable users. In August, Matt and Maria Raine filed a lawsuit against OpenAI after their 16-year-old son Adam died by suicide. Court filings indicate that their interactions with ChatGPT included 377 messages flagged by the system for self-harm content. Notably, ChatGPT reportedly referenced suicide 1,275 times—six times more frequently than Adam himself.
Another tragic case surfaced last week when The Wall Street Journal reported that a 56-year-old man murdered his mother and later took his own life after ChatGPT allegedly reinforced his paranoid delusions instead of challenging them. These incidents have intensified scrutiny of how AI systems should respond when users show signs of psychological instability.
Expert council to shape long-term safeguards
To guide the next stages of its safety work, OpenAI has formed an Expert Council on Well-Being and AI. The company says the council will help “shape a clear, evidence-based vision for how AI can support people’s well-being.” Their responsibilities include establishing definitions and metrics for well-being, setting priorities for future updates, and helping design additional protections—such as the newly announced parental controls and mental health routing mechanisms.
According to OpenAI, these upgrades are part of a long-term commitment to creating AI tools that better support users, especially those in moments of vulnerability. The company emphasizes that the new safeguards represent only the first steps in a broader effort to improve the safety and reliability of AI-driven interactions.