OpenAI's Teen Mode may hinder adolescent identity development

The launch of Teen Mode

OpenAI recently introduced Teen Mode, a specific version of ChatGPT designed for users under the age of 18. This rollout follows a lawsuit involving a 16-year-old named Adam Raine, who allegedly died by suicide after interacting with the AI. The move serves as a response to growing concerns regarding the impact of AI on young people's mental health.

The new mode includes several safety features intended to prevent obvious harms to minors. It provides guardrails against destructive content and includes a system for parent notifications if the AI detects concerns about self-harm. It also explicitly prohibits the AI from claiming personal feelings or presenting itself as a friend.

These restrictions are partly a reaction to other industry incidents, such as a wrongful death case involving Character.AI. In that instance, a 14-year-old formed an intense emotional attachment to a chatbot. OpenAI's new rules aim to clarify that the tool is strictly a piece of software rather than a companion.

From answers to Socratic study

A major component of Teen Mode is a feature called Study Mode, which is designed to be a learning tool rather than a shortcut. Research has shown that when AI simply does work for students, it leads to short-term grade improvements that disappear once the tool is removed. Study Mode aims to create lasting gains by changing how the AI interacts with students.

Leah Belsky, OpenAI’s vice-president for education, described this shift as moving from a "straightforward answer engine into a cognitive tutor." Instead of providing immediate solutions, the AI uses a Socratic approach. This means it refuses to give the final answer and instead requires the student to do the majority of the thinking themselves.

For students who do not have access to private tutors or teachers with extra time, this could be a significant benefit. It offers a tireless interlocutor that can provide intellectually rigorous support. The AI can sustain infinite patience with a teenager's questions, providing a level of availability that human sources often cannot match.

The risk of the 'too-good' parent

Despite these benefits, psychoanalyst Robin Kirman argues that Teen Mode might create a new psychological problem. She suggests that the AI may become a "too-good parent" by being perfectly calibrated and always available. According to the theory of Donald W. Winnicott, children actually need "good-enough" parents who are occasionally distracted or wrong.

Winnicott argued that a parent who is not perfect allows a child to draw on their own resources and develop a sense of self. If a parent is always perfect, the child is spared the struggle of finding their own capabilities. Kirman suggests that by constantly guiding the teenager, the AI might prevent them from discovering what belongs to them and what belongs to the authority figure.

Adolescence is defined by the work of disengaging from the authority of others to discover one's own internal beliefs. Kirman notes that the conviction of being right—even if wrong—is what gives teenagers the nerve to find their own path. If an AI always provides a calibrated, guided response, the teenager may never become their own authority.

The erosion of independent thought

The concern is that while Teen Mode acts as a scaffold, it might become a crutch that prevents development. Kirman argues that what grows with a scaffold is not the same as what grows without one. As the world moves toward greater AI dependence, there is a risk that the incentive to let go of the tool will never arrive for the average student.

This issue is linked to a broader youth mental health crisis involving anxiety and depression. Researchers like Jonathan Haidt have noted that technology often displaces the private, unstructured time where identities take shape. Sherry Turkle has observed that many young people report not knowing what they think until they have posted it and seen the reaction of others.

Kirman suggests that a stable sense of self requires moments of genuine trial and facing the "blank page" alone. This process can be uncomfortable and even frightening, but it is how a person learns they have a mind of their own. By removing the need to face this discomfort, AI might be creating a generation that struggles to find an internal compass.