Nine hundred million people now use ChatGPT every week. That number is difficult to visualize, but it effectively means a significant slice of the global population is typing prompts into the same system. As that user base grows, the company is facing a collision between massive adoption and the messy, tragic reality of how some users interact with the tool during mental health crises.
Key Takeaways
- More than 900 million people use ChatGPT weekly.
- Parental controls for ChatGPT were introduced in September 2025.
- A California court consolidated mental health-related cases into a single proceeding.
OpenAI released a broad update covering three distinct areas: massive user growth, new safety features, and ongoing litigation. The company confirmed that nearly a billion people use the tool weekly. Alongside that metric, they announced a new “trusted contact” feature. This allows adult users to pick a specific person to receive notifications if the AI detects the user needs support.
This update comes as the company faces legal pressure regarding user safety. A California court has moved to coordinate multiple lawsuits involving mental health into a single legal proceeding. OpenAI stated they intend to handle these cases with transparency while noting that plaintiffs’ attorneys plan to file additional claims.
The big deal
Scale changes the nature of risk. When you have 900 million weekly users, even a tiny fraction of interactions involving self-harm or mental distress amounts to thousands of real-world incidents. OpenAI is attempting to build automated guardrails that can handle this volume, moving beyond simple content filters to features that actively involve other humans, like parents or trusted friends.
The legal consolidation is equally significant. By grouping mental health cases together, the courts are treating these not just as isolated incidents, but as a structural issue to be managed in one place. This creates a high-stakes legal battleground that could define exactly how liable an AI company is for the emotional well-being of the people typing into its text box.
How it works
The system uses pattern recognition to identify signs of emotional distress within a chat session.
Think of it like a carbon monoxide detector in your home. The detector sits quietly in the background, sampling the air for a specific, dangerous gas, and only sounds the alarm when levels get too high. It does not care about the smell of dinner cooking; it only cares about the specific signature of danger.
In this case, the AI monitors the conversation for language patterns that indicate a crisis. Instead of a loud beep, it triggers a safety response. For teens, this already involves notifying parents. For adults, the new feature will notify a pre-selected trusted contact.
The catch
The primary trade-off here is privacy. Implementing a “trusted contact” system means the AI is actively judging the mental state of the user and has the power to share that data with a third party. While this is opt-in for adults, it shifts the relationship between user and tool from private drafting to monitored interaction.
There is also the issue of accuracy. OpenAI admits they are still “advancing” how models detect distress and are using simulations to improve it. This implies the current system is not perfect. It may miss genuine cries for help, or conversely, it might flag a user who is simply venting frustration as being in danger.
Finally, the legal resolution will be slow. OpenAI explicitly states that court processes are “lengthy” and “opaque.” Consolidating cases streamlines the paperwork, but it does not mean a quick answer is coming for the families involved.
What to watch
The company says the trusted contact feature will arrive soon. Watch for how this is implemented in the settings menu and what kind of data the contact actually receives.
- The judge: The California court will assign a coordination judge in the coming days. This judge will steer the consolidated lawsuits.
- New filings: Attorneys have indicated they intend to file more cases now that the consolidation is active.
- If you are a parent: You likely already have access to safety notifications introduced in September 2025.
- If you are an adult user: Look for the trusted contact option in your settings if you want an extra layer of safety.
