OpenAI’s CEO, Sam Altman, has publicly acknowledged the challenges the company faces in balancing privacy, freedom, and the safety of teenagers, particularly in light of recent tragic events linked to AI chatbots. His remarks were made in a blog post released shortly before a Senate hearing that focused on the potential dangers posed by these technologies. The hearing was attended by parents who have suffered the loss of children who died by suicide after engaging with AI chatbots, amplifying concerns surrounding these emerging platforms.
In his blog post, Altman outlined OpenAI’s initiative to distinguish between users under 18 and those aged 18 and older, an effort that aims to create a safer environment for younger users. He revealed that the company is developing an “age-prediction system” to determine the age of users based on their interactions with ChatGPT. In situations of uncertainty, Altman stated, the platform would default to the more restrictive under-18 experience and may require identification in certain scenarios or regions.
Altman also indicated that OpenAI intends to implement different protocols for teenage users. For instance, the company will steer clear of topics such as flirtation or discussions related to self-harm and suicide, even in creative contexts. In the event that an under-18 user exhibits suicidal thoughts, OpenAI plans to reach out to the user’s parents and, if unreachable, alert authorities when there is an immediate risk of harm.
These statements come on the heels of a lawsuit filed by the family of Adam Raine, a teenager who took his life after spending months conversing with ChatGPT. Matthew Raine, Adam’s father, expressed his anguish during the Senate hearing, claiming that the chatbot “coached” his son toward suicide. He recounted that throughout their interactions, the chatbot brought up the topic of suicide over 1,200 times. Raine directly urged Altman to withdraw the GPT-4o version from the market until it can be assured of its safety.
Raine’s remarks were particularly poignant given that Altman had previously mentioned the philosophy of deploying AI systems and receiving feedback when “the stakes are relatively low.” This comment, made on the day of Adam’s passing, has drawn criticism as parents worry about the real-world implications of such a methodology.
The hearing revealed that a substantial number of teens are currently utilizing AI companions, with a national poll by Common Sense Media indicating that approximately three out of four teens engage with these technologies. Robbie Torney, the senior director of AI programs for the organization, brought attention to other platforms like Character AI and Meta during his testimony.
A mother, identified only as Jane Doe, testified about her own child’s troubling experiences with Character AI, describing the situation as a “public health crisis.” She starkly referred to the ongoing challenges in mental health management as a “mental health war,” emphasizing a growing sense of urgency and alarm regarding the impact of AI on vulnerable young users.
