Social Media Child Safety Accord Reached; AI Protections May Lag

A significant accord reached by U.S. states this summer aims to enhance child safety on social media platforms, providing crucial lessons for the emerging challenges posed by artificial intelligence (AI). While measures to secure social media are still evolving, the approach taken offers insights into how to tackle the risks of AI for young users, a technology whose potential harms are just beginning to be understood.

Meta agreed to substantial changes and improved safeguards in an $18 billion settlement, acknowledging the known dangers of social media. However, advocates emphasize that AI's risks for children are becoming apparent and require swifter action than was initially taken with social media.

AI companies are starting to introduce safety features previously pioneered by social media firms, including age verification and restricted experiences for minors. Nevertheless, parents and experts stress that the primary lesson from the social media safety efforts is the necessity of proactive measures to mitigate AI's potential harms.

Concerns surrounding social media have largely focused on excessive screen time, its impact on mental health, data privacy, and risks like cyberbullying and online predators. AI presents potentially more pervasive issues, as children can interact with AI chatbots, use them for academic assistance, and even form unhealthy attachments. Experts are raising questions about how AI access affects children's cognitive development.

While AI offers undeniable benefits, such as tutoring and educational support, limiting its use presents unique challenges. Unlike social media, where time restrictions are more straightforward, applying similar limits to AI is complex. According to the Pew Research Center, 64% of U.S. teens have used AI chatbots, often for research and schoolwork, but also for entertainment and emotional support. Social psychologist Jonathan Haidt considers AI a greater threat to children than social media, warning of potential unhealthy attachments and "cognitive offloading."