Microsoft adopts binding AI safety and privacy principles for schools

Microsoft has agreed to ten safety and privacy principles for AI in schools, struck with the American Federation of Teachers (AFT), the second-largest teachers union in the United States, and its New York City affiliate, the United Federation of Teachers (UFT). According to The Verge, the terms are designed to be contractually enforceable by any school district that adopts them — a firmer footing than the voluntary pledges that typically govern AI products.

What the principles cover

The commitments include a pledge not to train AI models on student or educator data, limits on how much data Microsoft collects in the first place, and plain-language disclosures to families about how its tools work. The agreement also prohibits AI companions and requires human review of "high-risk" decisions.

Starting in November, The Verge reports, school districts will be able to attach the terms to new or existing Microsoft agreements without renegotiating their entire contracts — a mechanism intended to make adoption as frictionless as possible.

Why Microsoft moved now

The deal was announced a week after two of the largest US school systems, New York City and Los Angeles, put in place one-year bans on AI tools for many students, giving those districts time to weigh appropriate use and guardrails. With parents and some educators resisting AI and screens in classrooms more broadly, Microsoft appears to be bracing for similar decisions elsewhere.

A union that negotiates rather than only objects

The AFT has been sharply critical of technology in schools while staying open to working with tech companies on AI's arrival. In a May speech, AFT President Randi Weingarten called for banning screens before third grade and "student-facing AI" before middle school. Around the same time, the union announced a $23 million AI training hub for educators funded by Anthropic, Microsoft and OpenAI.

Commenting on the Microsoft agreement, Weingarten described the terms as "iron-clad" and said they were needed "because no one else, including the federal government, has stepped up to do the real work." She added: "We can get angrier and angrier, or we can act decisively; anything less than legally enforceable provisions is simply a wish list."

Why it matters

Most AI safety commitments from major vendors remain policy statements that customers cannot easily hold them to. This agreement converts privacy and safety promises into contract terms that school districts can enforce, and it hands every district in the country a ready-made template: from November, adopting the language requires no full contract renegotiation.

It is also a case study in how vendors are responding to local political pressure. With Weingarten arguing that federal action is absent, and with the biggest US districts willing to ban student-facing AI outright, binding contractual language becomes a way for a company like Microsoft to keep its tools in classrooms on terms educators can accept. If uptake is broad, the AFT–Microsoft deal could quietly become the de facto baseline for AI in US public education — and a precedent other unions and edtech buyers are likely to cite.