Given AI’s rapidly evolving and wide-reaching impact, policymakers have ongoing opportunities to mitigate risk, including to improve youth wellbeing. A tiered, risk-based approach—aligned with national and international standards for responsible AI use—could offer an adaptable and effective path forward.

Implications for Youth

Heightened safeguards may be particularly beneficial when a chatbot interacts with youth. Youth are growing up alongside rapidly evolving AI tools that increasingly shape how they communicate, learn, and form social connections, often overlapping with existing social media environments. While these tools can create new risks—such as amplifying bullying, reinforcing harmful social comparisons, or contributing to stress, anxiety, and isolation—they also hold promise when designed thoughtfully, including offering age-appropriate information, academic support, and early guidance toward trusted adults or resources.

Considerations for Oversight of AI Innovation 

Clear standards for AI development can help ensure that innovation supports healthy development, protects youth well-being, and maximizes the benefits of emerging technologies while minimizing unintended harm.

• Safeguards for youth
  • Age-gates, simplified language, human oversight when minors are involved, and heightened testing for school-age scenarios are all options that could mitigate risk of AI use for youth.
  • Chatbots could also be prohibited from generating or distributing sexually explicit, nonconsensual, or “morphed” images of children.

• Tiered approach
  • The EU AI Act requires higher obligations for systems that influence health decisions. It allows for lighter transparency for low-risk wellness chatbots.
    • This risk-based approach aims to balance innovation with safeguards for systems that may affect health outcomes.
  • Increased oversight could include mandating transparency and labeling for AI-generated content to improve accuracy and trustworthiness.
    • Public disclosure of training data sources can help users and regulators evaluate the reliability of AI outputs (e.g., systems trained primarily on opinion-based platforms may produce less evidence-based health information than systems trained on peer-reviewed or clinical sources).
    • Transparency and labeling requirements for AI-generated content could help users distinguish between human- and machine-produced information, improving trust and accountability.
  • Examples:
    • Prohibiting AI-powered robots or tools from being used to stalk, harass, or surveil individuals.
    • Clarifying the ownership and intellectual property rights of AI-generated content.
    • Establishing “Right to Compute” laws that set requirements for transparency and accountability in critical infrastructure controlled by AI systems.

• Independent conformity assessments
  • For higher-risk uses (e.g., addiction counseling), pre-deployment testing and ongoing audits against standardized evaluations could be required.
  • Agencies could conduct public, human-led impact assessments for AI systems that make consequential decisions, which supports transparency and accountability.

• Responsible innovation via states’ structural and economic strengths
  • States can explore public–private partnerships—e.g., venture funds or innovation hubs that support early-stage AI and quantum technology startups advancing public-interest goals.
  • Such efforts, combined with targeted research and development (R&D) incentives and collaboration among universities, community colleges, and private industry, can help states serve as leaders in safe, transparent, and socially beneficial technological development.

Strengthening AI Ecosystems for Public Benefit

Investing in talent pipelines and safe innovation environments not only strengthens AI development but also ensures tools are designed with youth wellbeing in mind.

• Regulatory Sandboxes
  • Create “regulatory sandboxes” or “learning laboratories” where companies can test new AI and quantum technologies under relaxed rules and with government oversight. This allows for rapid iteration and policy experimentation in a controlled environment.
• STEM Education and Talent Development
  • Support STEM education and research programs at state universities and colleges to build a local talent pipeline, including for quantum and AI research centers.
• Data and Infrastructure
  • Explore ways to expand the development of high-performance computing centers and data infrastructure, which are essential for AI training and quantum research.
• Standardization and Interoperability
  • Set AI priorities for state agencies with customized approaches for youth, health, and emergency use, while ensuring clear data protection and disclosure practices.
  • Work to align state-level policies with federal standards from organizations like NIST. This ensures that state efforts are compatible with national and international frameworks, reducing compliance burden and promoting interoperability.