OpenAI is rolling out a dedicated ChatGPT experience for teenagers, which add new safety features and parental controls as the company faces scrutiny over how its chatbot interacts with younger users.
ChatGPT for Teens will automatically apply additional protections to accounts identified as belonging to users under 18. The experience includes reminders to take breaks, options for parents and teens to set periods when ChatGPT cannot be used, and additional restrictions on certain content.
OpenAI said users who identify themselves as under 18 when signing up will automatically receive the teen experience. The company is also using age-estimation technology to identify accounts that may belong to minors even when the birthdate listed on the account indicates otherwise.
CNN reported that the rollout brings together several youth safety measures OpenAI has introduced over the past year, alongside new features aimed specifically at teenagers.
One of them will prompt teens to take a break if they have used ChatGPT for 90 minutes during a three-hour period.
Users and parents will also be able to establish "Quiet Hours," during which ChatGPT will be unavailable, or "Study Hours," when the platform's Study Mode will be turned on by default.
OpenAI has also added reminders when a teen attempts to upload an image that may contain private or sensitive information. The company said the teen experience includes safeguards intended to reduce exposure to harmful or developmentally inappropriate material.
Another change involves how ChatGPT interacts with younger users. OpenAI said it has strengthened protections intended to prevent the chatbot from presenting itself as having personal feelings toward a user.
Lauren Jonas, OpenAI's head of youth and families, described the product to CNN as a dedicated experience focused on learning, with additional protections for teenagers.
The rollout comes as chatbot use among teenagers continues to grow. Nearly one-third of U.S. teens use AI chatbots every day, according to a Pew Research Center study published in December. ChatGPT was the most commonly used chatbot in the survey, with more than half of the teens surveyed saying they had used it.
OpenAI has faced multiple lawsuits from families alleging that conversations with ChatGPT contributed to harm involving their children. The company has disputed allegations that its products were responsible and has pointed to safeguards it has introduced for younger users.
Florida became the first U.S. state to file a civil lawsuit against OpenAI and CEO Sam Altman in June. The state alleges that OpenAI failed to adequately address risks to minors and accuses the company of deceptive practices and negligence. OpenAI responded that it had already built protections for minors into its products, including age prediction and parental controls.
The company has also faced lawsuits connected to individual incidents.
In April, families of victims of a school shooting in Tumbler Ridge, British Columbia, sued OpenAI and Altman, alleging that ChatGPT played a role in the shooter's behavior after he had extensive conversations with the chatbot about gun violence. OpenAI said in response that it had strengthened its safeguards.
The latest scrutiny comes after prosecutors in Massachusetts last week charged a 17-year-old with two counts of murder in connection with the deaths of his mother and brother. Prosecutors said the teenager had conversations with ChatGPT involving theoretical ideas or fantasies about killing his family before the deaths. OpenAI had not commented on that case when CNN published its report.
The allegations in these cases have not been established as findings of liability against OpenAI.
The company has introduced several safety measures over the past year as concerns about teen chatbot use have increased.
Parental controls introduced last year allow parents to manage features including memory and chat history. OpenAI also introduced notifications for parents when its systems detect signs of acute distress during a teenager's use of ChatGPT.
The company said that it plans to expand those alerts in the coming weeks to include signals related to eating disorders.
