OpenAI's new ChatGPT for Teens includes content restrictions and homework safeguards, but experts warn that limiting access entirely may be the only way to protect adolescents' developing brains from AI's cognitive and social impacts. The platform, designed for ages 13 to 17, promises to block sexual, violent, and self-harm content while steering students toward learning rather than providing direct answers. Yet emerging research suggests these guardrails address only surface-level concerns, missing deeper developmental risks.

What Does the Research Actually Show About AI and Teen Brains?

Recent studies paint a troubling picture of how AI use affects adolescent cognition. A study of 27,000 students ages 12 to 18 conducted by researchers at the University of Hong Kong and Stockholm University found that after six months of using AI, students saw their average homework scores improve. However, when exam time arrived, those same students scored 20 percent below their classmates who had not used AI tools. This gap suggests that while AI can produce short-term homework gains, it may undermine the deeper learning and retention needed for long-term academic success.

A 2026 literature review of 400 studies by the Brookings Institution concluded that "the risks of utilizing generative AI in children's education overshadow its benefits". Beyond academics, the social and emotional toll is equally concerning. A 2025 survey by Thorn, a child safety nonprofit, found that nearly 29 percent of minors had turned to AI chatbots for guidance on sensitive or intimate topics, bypassing conversations with parents and peers. This shift represents a fundamental change in how adolescents seek support during critical developmental years.

Are Content Filters and Parental Controls Actually Effective?

While 80 percent of American parents support more AI controls for children, including parental permissions and alerts, experts question whether these safeguards will work in practice. The history of age-restricted technology offers little reason for optimism. Social media platforms have struggled for years to enforce age controls, and ChatGPT for Teens has no legal force behind age verification. More troubling, a teen-specific version of the app may actually increase adolescent AI use by making parents and educators feel more comfortable with access.

The problem extends beyond OpenAI's platform. A Common Sense Media survey found that 63 percent of teens are explicitly using AI to get homework answers. When researchers tested Google Search's AI assistant, it completed every math assignment and essay they asked it to solve. Critically, there is no way to turn Google's AI assistant off, meaning even schools and parents who want to restrict AI access cannot fully prevent exposure.

How to Protect Teens From AI's Developmental Risks

• School-Based Strategies: Return to pen-and-pencil exams and paper-based homework to eliminate AI access during critical learning moments, ensuring students develop problem-solving skills without AI assistance.
• Device-Level Controls: Require technology companies to grant parents and teachers the ability to block AI on devices used by children, moving beyond app-level restrictions to hardware-level enforcement.
• Home Environment Changes: Delay personal screen use as long as possible and prioritize offline activities like reading, outdoor time, and face-to-face social interaction to support healthy cognitive and social development.
• Skepticism Toward Teen-Focused Apps: Approach apps specifically designed for children with particular caution, recognizing that marketing safety features may create false confidence in tools that still carry developmental risks.

The challenge facing parents, educators, and policymakers is that partial measures may backfire. Guardrails on ChatGPT for Teens could create a false sense of security, leading more families to embrace AI use without understanding the underlying risks. Meanwhile, the same tools remain available elsewhere without restrictions, and competing platforms like Google's AI assistant offer no opt-out mechanism.

The pattern is familiar. Over the past decade, families have navigated smartphones, social media, and screen-based schooling, each time discovering after-the-fact that early adoption carried unforeseen costs. With AI, the stakes feel higher because the technology is advancing faster and the research on long-term effects remains preliminary. Yet the early evidence suggests that cognitive decline and social friction are not edge cases but predictable outcomes of heavy adolescent AI use.

OpenAI's effort to add safety features to ChatGPT for Teens represents progress in acknowledging the problem. But if the goal is truly to protect adolescent development during this critical window, experts argue that limiting access entirely, rather than making access safer, may be the only strategy with proven results. The question now is whether schools, parents, and policymakers will act on that insight before another generation of data confirms what common sense already suggests: developing brains and addictive technology are a dangerous combination.