Advancing Safer AI Standards for Children and Mental Health

Spring Health welcomes the U.S. House of Representatives’ passage of the Kids Internet and Digital Safety legislation, including the Safeguarding Adolescents From Exploitative BOTs Act, as an important step toward stronger safeguards for children online.

As AI chatbots become more common in emotional support and mental health-related interactions, clear safety standards and transparent disclosure are essential. Families, policymakers, and the public should be able to understand whether these systems have been evaluated against rigorous standards, especially when conversations may involve emotional distress or mental health risk.

Spring Health has focused on this need through VERA-MH, the first open-source, clinically validated AI safety benchmark for mental health. Spring Health co-developed and open-sourced VERA-MH with the AI in Mental Health Safety and Ethics Council to give the field a transparent, evidence-based way to evaluate how AI systems respond in high-risk mental health contexts.

The current iteration of VERA-MH focuses on adult suicide-risk conversations. It evaluates whether AI systems can detect potential risk, confirm risk when needed, guide people to human care, hold a supportive conversation, and stay within appropriate AI boundaries. That scope matters. Youth-specific AI safety standards will require their own dedicated evaluation, research, and clinical input.

As the legislation moves to the Senate, Spring Health encourages policymakers to strengthen disclosure requirements so people can see whether AI systems have been evaluated against clinically grounded safety standards. VERA-MH offers one model for how transparent, evidence-based evaluation can help make AI safety more specific, measurable, and accountable.

Spring Health stands ready to serve as a resource to policymakers as they refine approaches to AI in mental health and work toward safer digital environments for children and families.

‍