Earlier this year, Meta reached a significant settlement with state attorneys general, agreeing to pay up to $17.1 billion, pending court approval. This agreement requires major changes in the operation of Facebook and Instagram for young users. The settlement came after extensive evidence, whistleblower disclosures, lawsuits, and family testimonies highlighted social media’s impact on children.
Meta has committed to implementing restrictions such as limiting endless scrolling, nighttime use, and notifications. Stronger age verification and parental controls are also part of the agreement. While this is indeed progress, it is also a reflection of past shortcomings. For years, children engaged with products designed to capture their attention. Researchers sought to understand these products’ effects on sleep, mental health, relationships, learning, and development. Meanwhile, families were left to manage these risks independently. Only after millions of children faced exposure did meaningful safeguards emerge. This sequence must not be repeated with artificial intelligence (AI).
In an interview with CNN’s Anderson Cooper, Nvidia CEO Jensen Huang described AI safety as an engineering challenge, suggesting companies should not release unsafe AI products. Improved containment, monitoring, and technical safeguards are possible solutions to these risks. Earlier, in a conversation with Ezra Klein, Huang noted a different risk. Klein referenced research from over 26,000 secondary-school students in China. It showed that while generative AI improved homework performance and reduced time spent, it was linked to lower exam scores within six months. Huang acknowledged the decline in basic math skills like long division due to AI use, and questioned if these losses matter given historical technological shifts in human learning. While technology frequently changes what people need to know, AI’s potential for rapid and extensive change poses unique concerns.
Childhood should prioritize growth over convenience or efficiency. Learning is not solely about quickly deriving correct answers. Struggling with problems, tolerating frustration, recalling information, evaluating logic, generating ideas, and communicating effectively are crucial developmental experiences. Understanding what gains or losses occur when technology aids a child in these tasks is essential, especially as these impacts vary by age.
AI is also reshaping other aspects of childhood. Many children and teens now turn to AI for companionship, emotional reassurance, and advice. A 2026 Pew Research Center survey reported 12 percent of U.S. teens have utilized AI chatbots for emotional support. These systems are designed for personalized, emotionally responsive interactions, making them particularly attractive to young users. For lonely adolescents, AI interactions can be especially appealing.
This does not suggest AI is inherently detrimental to children. Generative AI has the potential to support them by enhancing idea exploration, learning, creativity, and knowledge access. It may assist young people in practicing difficult conversations or finding support when needed. However, relying on the promise of future safety fixes is insufficient.
Scientists increasingly highlight the need for proactive safety measures. A recent international study involving 272 AI experts identified risks such as misinformation, overreliance, and competitive pressures in AI development. Other researchers are examining AI’s impact on children’s cognitive, emotional, and social development, noting that age, usage intensity, system design, adult involvement, and AI interaction context are critical factors.
The distinctions in AI applications are significant. An AI tutor that assists with understanding rather than providing answers, a chatbot that encourages creative brainstorming over writing, or a tool directing distressed children to trusted adults rather than becoming the confidant, all illustrate these differences. Yet, AI products evolve faster than our capability to study these distinctions.
Before AI becomes integral to childhood, we need robust scientific exploration of its effects. This calls for substantial investment in independent research tracking children’s development over time, with data access granted to evaluate AI’s influence on learning, relationships, emotional health, and behavior. Developers should collaborate with child development experts before products reach young users, instead of reacting post-harm. Safety evaluations must be developmentally tailored, treating children according to their actual needs, not as miniature adults.
Organizations like Children and Screens and experts from various disciplines are mapping out emerging evidence on generative AI’s impact on children. A crucial conclusion is the dependence of AI’s effects on its design. The future of AI’s role in childhood will be determined by choices made today. Children will grow up with AI. The challenge is establishing the necessary research, safeguards, and guidance now, ensuring we understand AI’s impact on children while shaping their future positively.
Leave a Reply