Microsoft and teachers' unions unveil enforceable 'National AI Safety & Privacy Standard' for schools

A Microsoft news release dated September 9, 2026 in New York announces a National AI Safety & Privacy Standard for schools, agreed by Microsoft, the American Federation of Teachers and the United Federation of Teachers. Randi Weingarten of the AFT, Michael Mulgrew of the UFT and Microsoft's Brad Smith presented it after months of negotiation. They describe it as legally enforceable protection for students, families and educators, prompted by a lack of meaningful federal and state rules. The release calls it a first of its kind and links it to New York City Public Schools' recent ban on screens and AI.

The standard is meant to bar technology companies from using student data to train AI, to rule out tracking of students, to prevent AI from deciding without human oversight, and to oblige companies to give educators and parents plain-language answers about how tools work. School districts anywhere in the US can write the protections into their Microsoft customer agreements, which makes them enforceable by contract. Smith said Microsoft intends to take the agreement to districts nationwide.

The release organizes the commitments under three headings. On privacy, information about students and educators cannot train models or be sold or reused, and schools decide how data is retained and erased. On safety, systems must be built to avoid harmful or manipulative experiences, with strong security and human oversight. On transparency and control, families get clear information on AI use and data collection, schools choose whether and in what way AI is deployed, and vendors must be upfront about product changes and risks. Mulgrew said districts can end agreements and seek damages if companies break the rules. Weingarten noted that similar frameworks were adopted elsewhere, including Los Angeles Unified.