AI products need a key benchmark: children's safety
Our approach to evaluating AI is modeled on the independent crash-test ratings for vehicles that contribute to safer cars. We set clear standards for child safety, test the AI products children use most and share the results to hold the industry accountable.
Our Latest Assessment
AI Mental Health Apps
The market for AI mental health apps is unregulated, unstable, and in some cases actively harmful to teens.
Risk Assessments
Since 2024, Common Sense Media has conducted independent human evaluations of safety, effectiveness, and appropriateness for AI systems and products that are used by kids, teens, and in school. See our recent risk assessments:
Methodology
How we set standards and conduct evaluations
Our methodology layers expert standards, human red-teaming, and automated evaluation.
On an ongoing basis, we consult with subject-matter experts in child development, mental health, and related fields to develop and publish clear standards.
Our product evaluations start with red-teaming: human experts stress-test tools in realistic, adversarial, multi-turn, and youth-specific scenarios. As we validate these frameworks, we translate them into automated workflows that run alongside our human testers.
We also partner with technical evaluation and infrastructure organizations — including Transluce and Humane Intelligence — to automate our standards and rubrics and scale our evaluations.
Get Involved
The choices we all make in the next few years will shape how kids learn, think, and relate to the world.
Stay connected
Stay connected
Sign up for emails to support and follow the work as it develops, joining the community of people who care about getting this right for kids.