OpenAI has just announced the global rollout of a version of ChatGPT specifically designed for teenagers aged 13 to 17. This update introduces targeted technical restrictions, notably regarding image generation and access to sensitive content such as violence, eating disorders, self-harm, or sexually explicit material. In addition, the company is strengthening its parental alert system and introducing new features like "study mode," which prioritizes step-by-step reasoning over immediate answers to mitigate the risk of "cognitive offloading" among young users.
This initiative comes amid a climate of legal tension marked by several lawsuits filed by families. Some cases, brought before Californian and Canadian courts, accuse the platform of failing to prevent tragedies—specifically by providing inappropriate responses to minors in psychological distress or by failing to alert authorities to clear threats of violence. These cases raise fundamental questions regarding the legal liability of AI developers when it comes to the risky behavior of their youngest users.
Beyond safety, the tech giant, valued at tens of billions of dollars, must navigate an increasingly restrictive regulatory landscape. As many states consider strict limitations on minors' access to digital platforms, OpenAI is looking to get ahead of potential regulations while safeguarding its brand image. This strategy is all the more crucial as the company actively prepares for its IPO, a major milestone where the management of ethical risks and the protection of minors will be under close scrutiny by future investors.
The success of this deployment now rests on two pillars: the actual effectiveness of age-verification tools and the AI's ability to detect risky behavior in real time. While OpenAI asserts it wants to balance teenagers' right to privacy with their safety, the challenge remains immense. The firm is banking on a cautious, evolving approach to reconcile the educational use of its tools with mental health imperatives, all while awaiting court rulings that will ultimately define the legal framework of liability for future conversational models.