Microsoft, AFT Set Enforceable AI Guardrails for Schools
Microsoft, AFT Set Enforceable AI Guardrails for Schools
Anonymous
Key Points
1
The agreement is structured as a binding Memorandum of Agreement between the National Academy for AI Instruction and Microsoft, with provisions for additional AI providers to sign on. AFT President Randi Weingarten and Microsoft President Brad Smith signed it on Sept. 7, two days before the public announcement.
2
Microsoft says the arrangement includes auditing provisions, with audits conducted at Microsoft and the results made available to stakeholders—an accountability mechanism beyond merely publishing voluntary principles.
3
New York City’s restrictions include limits on one-on-one screen time: for students in third grade through eighth grade, schools are advised to cap individual screen use at 30 to 45 minutes per day, in addition to the city’s restrictions on student-facing AI.
4
The agreement includes commitments governing the export and deletion of school data, alongside its provisions on profiling, biometric information and AI memory features.
5
AFT President Randi Weingarten had previously called for a total screen ban from kindergarten through second grade and a ban on student-facing AI in elementary school, linking the agreement to a broader campaign over children’s device and online safety.
Microsoft and two major teachers unions announced a binding AI safety standard for U.S. schools on September 9, creating Yahoo News enforceable legal protections for student and educator data. Starting November 1, the agreement bars Microsoft from using school data to train AI models or sell it for advertising and other products, except for narrow safety exceptions.
The deal comes as New York City and Los Angeles have restricted student AI use amid limited federal oversight. Yahoo News reported the agreement includes human review of major decisions, bans on tracking and AI companions, plain-language disclosures for families, and 72-hour breach reporting requirements.
How Microsoft Will Protect Student Data
Microsoft agreed not to train its AI models on student or educator information, Unite AI reported. The company also cannot sell the data, use it for ads, or develop other products with it. The only exception is safety-related uses—like detecting harmful content or abuse.
The standard requires plain-language privacy notices for families and educators, Yahoo News stated. It also mandates data breach notifications within 72 hours and controls on how schools export or delete their information.
Audits and Accountability Beyond Voluntary Pledges
The agreement includes actual auditing provisions—a step beyond typical corporate promises. Yahoo News noted that Microsoft will conduct audits at its facilities and share results with stakeholders. This creates real accountability, not just voluntary compliance.
AFT President Randi Weingarten and Microsoft President Brad Smith signed the Memorandum of Agreement on September 7. The deal is structured so other AI providers can adopt the same standard, Yahoo News reported.
What Schools Still Need to Figure Out
The agreement applies to all school districts with Microsoft contracts starting November 1—no opt-in required. But districts must still determine which specific products are covered and identify all third-party data processors involved.
Schools also need clarity on enforcement mechanisms if breaches occur or safety exceptions are claimed. Yahoo News emphasized this is not a federal requirement or industry-wide standard, so each district must review its own vendor agreements.
New York and Los Angeles Crack Down on Classroom AI
New York City restricted student-facing AI and screen time limits for elementary students. Yahoo News reported schools are advised to cap individual screen use at 30 to 45 minutes daily for third through eighth graders. Los Angeles imposed similar restrictions.
AFT President Weingarten previously called for a total screen ban in kindergarten through second grade and a ban on student-facing AI in elementary school. The Microsoft agreement reflects her broader push for safer device and online practices for children.
Sentiment
Mostly NegativeNEGATIVENEUTRALPOSITIVE
Publishers
14
Articles
7
Reach
21
Topics
A
AnonymousAuthor
