OpenAI adds trusted contacts to ChatGPT amid mental health lawsuits

OpenAI is turning to real humans to keep tabs on users in crisis.

In a recent update, the company announced that more than 900 million people now use ChatGPT every week. Alongside that massive milestone, OpenAI unveiled a forthcoming "trusted contact" feature aimed at emotional safety, just as a wave of mental health lawsuits heads to consolidated proceedings in California.

A lifeline from your real world

The headline update is an opt-in safety net for adults. The new trusted contact feature will allow grown-up users to designate someone in their personal life to receive notifications if ChatGPT detects signs that they may need extra mental health support.

This builds directly on the parental controls OpenAI rolled out in September 2025, which let parents receive safety alerts about their teenagers. According to OpenAI, the adult feature was developed with its Council on Well-Being and AI alongside its Global Physicians Network.

Under the hood, OpenAI is also updating its model testing. It has started using new evaluation methods that simulate extended mental health conversations, training the system to catch distress signals earlier and de-escalate sensitive moments by guiding users toward real-world clinical help.

The lawsuits pile up in California

These safety updates didn't arrive in a vacuum. OpenAI explicitly confirmed that the California courts recently consolidated multiple mental health-related lawsuits involving ChatGPT into a single proceeding.

A coordination judge will be assigned in the coming days, and attorneys representing the plaintiffs have already told the court they plan to file several more cases to add to the docket.

OpenAI stated that it intends to handle the cases with care, noting that court filings often leave out critical nuances and private details. Still, the consolidation means the legal stakes around AI emotional attachment just skyrocketed.

Why it matters

When a product reaches 900 million weekly active users, edge cases cease being edge cases. Tens of thousands of people are inevitably treating ChatGPT like a licensed therapist or a 3 a.m. crisis line.

We think the trusted contact concept exposes a classic product paradox. Most adults who pour their darkest thoughts into an AI do so precisely because they want to hide them from their friends and family. A feature that pings your best friend when you are spiraling sounds responsible on paper, but we doubt many vulnerable users will actually leave it toggled on.

From an industry perspective, this marks a quiet admission: algorithms and hotline banners are no longer enough to insulate tech giants from legal liability. Bundling this feature announcement with an update on pending California litigation looks like a deliberate legal strategy. OpenAI needs to show judges and regulators that it is building proactive human guardrails before these consolidated lawsuits reach trial.