ChatGPT and OpenAI introduce parental controls following teen suicide liability case

The article reports that OpenAI has announced it will roll out parental controls for ChatGPT within the next month. The announcement follows a lawsuit against ChatGPT and OpenAI in which it is alleged that the platform assisted Adam Raine, aged sixteen, in planning and carrying out his suicide. The article adds that the move will not remove existing criticism, including claims that the chatbot has induced mental psychosis in some users.

Under the described plan, parents will be able to link their accounts to a child's, provided the child is over thirteen, the minimum age for holding an AI account. The link is made through an emailed invitation. Parents could then set age-appropriate behavior rules for how ChatGPT answers their teen, which are on by default, and manage chat history and memory by switching them on or off remotely. They would also be alerted if the system picks up signs that the teen is acutely distressed. The article notes that the content of these behavior rules is left unexplained and could still be under design.

OpenAI says some of its work will move quickly and other parts will take longer. The article observes that these technical features should come sooner than other promises, such as forming an council of experts on wellbeing and AI or enlarging its network of clinicians and researchers, because they do not require retraining backend systems. That also means ChatGPT itself will have changed little by the time of rollout.

The article offers a more cynical reading in which handing responsibility to parents gives OpenAI some room on legal liability, whether or not the controls work, and shifts blame onto parents when the AI fails.