OpenAI on Thursday announced the introduction of Trusted Contact, a new safety feature designed to alert a trusted third-party, such as a friend, family member, or caregiver, when an adult engages ChatGPT in a conversation about self-harm.

Trusted Contact joins the existing safeguard measures on ChatGPT as OpenAI battles with a series of lawsuits from families who said ChatGPT aided their loved ones in taking their own lives.

Adults over the age of 18 can add another adult they trust as a Trusted Contact from their ChatGPT settings. The Trusted Contact will receive an invitation explaining why they have been added, and must either choose to accept or reject the invitation within a week. Users can always add another person if the initial person they chose declines.

When users engage in a self-harm conversation that indicates a serious safety concern, ChatGPT’s automated monitoring systems detect it and notify the user that their Trusted Contact may be alerted. The system also encourages the user to reach out to their Trusted Contact with programmed conversation starters. OpenAI says every system notification undergoes a trained human review in under an hour before the Trusted Contact is alerted about the situation via email, text message, or an in-app notification if they already have a ChatGPT account.

Trusted Contact builds on the existing parental controls notifications that send alerts to parents or guardians when a linked teen account exhibits signs of “acute distress.” ChatGPT is also trained to direct people with suicidal tendencies to seek professional help.

OpenAI says that Trusted Contact is part of its “broader effort to build AI systems that help people during difficult moments,” in a blog post. “We will continue to work with clinicians, researchers, and policymakers to improve how AI systems respond when people may be experiencing distress.”