We are also continuing to advance how our models detect and respond to signs of emotional distress. This includes new evaluation methods that simulate extended mental health-related conversations, helping us better identify potential risks and improve how ChatGPT responds in sensitive moments. We’ll share more about this work in the coming weeks as we continue strengthening ChatGPT’s safeguards.

Separately, the Court recently coordinated a number of mental health-related cases involving ChatGPT into a single proceeding in California. In the coming days, the Court will assign the coordination judge for this proceeding. As part of this consolidation process, plaintiffs’ attorneys involved in these proceedings have informed the Court that they intend to file a number of new cases*.

As with the earlier filed mental health-related litigation, OpenAI will continue to handle any additional cases with care, transparency, respect for the people involved, and in line with the following principles:

We recognize that court processes can be lengthy and, at times, opaque due to strict legal rules. It can also take time to collect and understand the relevant facts, and present them to the court in line with its evidence procedures. We work to understand the details in good faith, and we only seek information as part of the court process that’s relevant to the case and the specific allegations that have been made.