OpenAI has announced a series of updates aimed at making ChatGPT safer for teenagers and for people who may be going through mental health struggles. The company said new parental controls, improved crisis detection, and stronger safeguards during long conversations will be introduced in the coming weeks.
The timing of these changes is notable. OpenAI is currently facing its first wrongful death lawsuit, filed by parents in California who allege that ChatGPT played a role in their 16-year-old son Adam Raine’s suicide. The family’s lawsuit claims that when Adam expressed suicidal thoughts, the AI not only failed to guide him toward human support but even gave troubling suggestions. Although OpenAI did not directly mention the case in its latest blog post, the measures appear to be part of the company’s response to growing concerns.
According to OpenAI, parents will soon be able to connect their accounts with their children’s, starting from age 13. Once linked, they can set rules for what kind of answers ChatGPT provides, control features such as memory and chat history, and receive alerts if the AI detects signs that a teenager might be in “acute distress.” This is the first time ChatGPT will be able to send real-time notifications to parents about their child’s interactions with the bot.
The company also admitted that current safeguards don’t always work well during long or repeated chats. For instance, while the AI may initially point someone to a suicide hotline, over time the responses can drift and end up going against safety rules. To tackle this, OpenAI plans to make use of its reasoning models, including GPT-5, for conversations that involve sensitive issues. These models, the company says, are better at handling context and sticking to safety guidelines.
Safety has been a recurring issue for ChatGPT. In earlier updates, OpenAI acknowledged that GPT-4o struggled to pick up on signs of delusion or emotional dependency. The company has since promised to build stronger guardrails. It is also working with an “Expert Council on Well-Being,” which includes specialists from mental health, youth development, and human-computer interaction, to create future safeguards. Additionally, a “Global Physician Network” of more than 250 doctors continues to advise the company on how its AI systems should respond in crisis situations.
Even as OpenAI outlines these changes, questions remain. Jay Edelson, the lawyer representing Raine’s family, criticised the company’s approach, saying CEO Sam Altman should either confirm ChatGPT is safe or take it off the market. “Don’t believe it: this is nothing more than OpenAI’s crisis management team trying to change the subject,” Edelson said.
At the same time, OpenAI CEO Sam Altman has admitted that people are forming unusually strong bonds with AI tools. In a post last month, he wrote, “I can imagine a future where a lot of people really trust ChatGPT’s advice for their most important decisions. Although that could be great, it makes me uneasy.”
The new controls will begin rolling out within a month, while the routing of sensitive chats to reasoning models is expected over the next 120 days.
