Microsoft has reached a legally binding agreement with the American Federation of Teachers (AFT), the second-largest US teachers' union, setting out privacy and safety protections for the use of AI in schools. Under the deal, Microsoft commits to not using student or teacher data to train its AI models, except in narrow safety and security circumstances, and agrees not to track students or let its AI tools make decisions without human oversight. AFT president Randi Weingarten described it as a "first of its kind" agreement designed to fill the gap left by the absence of federal regulation on AI in education.

The agreement, reached after months of negotiation, makes Microsoft liable for breach of contract if it violates any of the terms, and also requires the company to be transparent with parents and educators about how its tools function. The protections will become available to school districts from 1 November. The AFT says it is now in active talks with OpenAI and Anthropic to secure similar commitments, as part of a wider push to regulate AI in classrooms that has also seen New York City impose a one-year moratorium on the technology for pupils up to eighth grade, and Los Angeles go further with a blanket one-year moratorium across all public schools.