Claude Mythos and Safe/Effective AI Tutoring

The huge potential benefits of daily student-AI tutoring are matched by massive risks. Learners, parents, and educators all want to know if AI is both safe and effective. The time to act to create safety and efficacy assurances for minors is now. Government AI controls only become bipartisan when focused on minors, but could become applicable for all ages to opt in.

Ed3/EDU, A4L, and FullScale will convene two panels of experts to create and oversee a framework that evaluates evidence from providers on the safety and efficacy of their AI tutors. We will convene a national summit. In January 2027, we will publish the first “consumer reports” of safe and effective AI learning. District and family consumers and state regulators can use these reports to preference or restrict technologies.