OpenAI Calls for International Youth AI Safety Institute Ahead of G7 Summit

OpenAI has called for the establishment of an international youth safety institute to advance global standards for artificial intelligence use by young people. The announcement comes ahead of the G7 Leaders' Summit in Évian, France, where youth AI safety is expected to be a key focus.

The company argues that while AI offers significant opportunities for personalized learning and skill development, ensuring safe and age-appropriate access should not fall primarily on parents or young people themselves. Instead, companies should build products with appropriate safeguards by default while empowering families with tools and information.

OpenAI proposes that a dedicated institute—whether newly created or housed within an existing national AI institute with an expanded global mandate—would provide continuity beyond a single summit. Such an organization would help governments, researchers, civil society, and industry share evidence, develop practical guidance, and raise standards over time. The initiative could build on existing efforts, including Common Sense Media's Youth AI Safety Institute (supported by the OpenAI Foundation), collaborations with educators such as the American Federation of Teachers, and real-world deployments like Estonia's national ChatGPT rollout in schools, which OpenAI is studying alongside Stanford and Estonian researchers.

The company outlined nine principles that it believes should guide strong AI youth safety frameworks:

1. Age verification and default protections: Companies should know when a user is a minor and apply age-appropriate protections, using privacy-preserving age estimation and defaulting to safeguards when age cannot be determined.
2. Proactive risk and benefit assessments: Providers should complete annual youth safety risk assessments and implement proportionate safeguards before harm occurs, considering developmental stages, empirical evidence, and positive outcomes such as learning and creativity.
3. Accessible parental controls: Parents and guardians should have easy-to-use tools to manage settings such as memory, data use, and time limits, with active promotion of these features.
4. Transparency: Families deserve clear information about safeguards, parental tools, benefits, and risks, with published safety policies that are updated as risks evolve.
5. Protocols for high-risk situations: Companies should have clear procedures to address serious safety issues including self-harm, exploitation, grooming, and sexually exploitative content, including in-service support, referrals, and timely parental notifications. Systems should also be designed to prevent generating graphic sexual or violent material unsafe for children.
6. Support for real-world development: AI systems should support learning, development, and real-world relationships rather than replace them, with clear boundaries where human judgment and professional support are essential.
7. Privacy protections: The personal information of minors should be protected, with prohibitions on privacy-invasive targeted advertising and the sale of personal information.
8. Opportunity and literacy: Frameworks should promote access to AI tools that support learning and creativity while equipping young people with critical-thinking skills and AI literacy.
9. Accountability and audits: Strong accountability mechanisms, including independent audits under common standards and legislative oversight, are essential to ensure protections are meaningful.

OpenAI stated that these principles are reflected in how it builds and operates ChatGPT. The company has strengthened safeguards for users under 18, launched parental controls with proactive notifications, and advanced age-prediction systems to apply stronger protections when someone may be under 18. Its Model Spec includes dedicated principles for users under 18, prioritizing teen safety and encouraging trusted offline support when needed.

Through its Education for Countries program, OpenAI is working with governments and educators on research-driven deployments, localized learning tools, and teacher training. Partner countries include Estonia, Greece, and Singapore.

Separately, OpenAI will bring the OpenAI Forum to Paris to convene governments, researchers, civil society, educators, and industry. France's Ambassador for AI and Digital Affairs Clara Chappaz, OpenAI Chief Global Affairs Officer Chris Lehane, and youth safety leaders from iRaise/Everyone.AI are expected to discuss how young people can benefit from AI while reducing risks.

Source: OpenAI Blog