Microsoft’s Student AI Privacy Deal Pressures Google On School Data Rules

AP, carried by ETTech, reported that Microsoft has accepted legally binding privacy and safety terms for artificial intelligence used in US schools, creating an early test of whether a voluntary pact can influence how large technology vendors handle children’s data.

The deal links Microsoft with the American Federation of Teachers, a major US teachers union, at a moment when districts are slowing down classroom AI adoption rather than treating new tools as ordinary software upgrades.

New York City and Los Angeles, the two largest US school districts, have both imposed one-year pauses on student AI use while they review classroom products, contracts, privacy safeguards, transparency and accountability.

Microsoft’s school AI products now carry several data-use limits.

Information from students and educators is off limits for model training except in a narrow safety context.

The company also accepted restrictions that prevent collected school data from being monetized, folded into advertising systems or reused as product-development material.

The agreement also targets product design.

Tools may not include companion-style functions that are built to create emotional reliance, and classroom features are not supposed to stretch student attention beyond the work needed for a learning task.

Families should receive plain-language explanations of how Microsoft’s tools operate, with outside audits checking whether the commitments are being followed.

Brad Smith, Microsoft’s vice chair and president, said the standards will cover every district with a Microsoft contract beginning Nov. 1.

That date turns the framework from a public promise into an implementation deadline for schools already using the company’s education technology.

AFT President Randi Weingarten described the arrangement as a first-of-its-kind step toward a broader industry benchmark while lawmakers have yet to set comprehensive AI rules for schools.

She also contacted Google before the announcement, hoping the company would join the same framework.

Google’s response matters because its classroom footprint is much larger than a single product launch.

Google Classroom serves 150 million teachers and students globally, and the company faced criticism in August after enabling Gemini access for many K-12 users through its learning-management system.

The rollout applies only where schools permit Gemini or Gemini Notebook, but it still moved beyond Google’s earlier position of limiting Gemini access to students who were at least 18.

OpenAI and Anthropic are also negotiating with the AFT.

OpenAI framed the Microsoft pact as a milestone for school AI safety and privacy and said it was working toward its own agreement.

Anthropic said it was helping shape a safety and privacy standard with the union.

The policy gap remains unresolved.

Josh Golin, executive director of Fairplay, warned that schools could mistake a data-protection deal for proof that AI tools belong in classrooms.

His concern is that questions about age, purpose and educational value may be pushed aside if vendors can point to a privacy checklist.

Elizabeth Laird, a director at the Center for Democracy and Technology, saw a different practical effect: common standards could reduce the need for individual school districts to bargain alone with powerful technology companies.

Microsoft’s limits also reach data categories such as gender, race, age and ZIP code, which can help identify a person even after names are stripped away.

For schools, the agreement moves part of the privacy burden from districts and parents to the vendor.

Its policy value now depends on compliance audits at Microsoft and on whether Google, OpenAI, Anthropic and smaller education technology providers adopt comparable limits before fragmented local contracting becomes the default rulebook.