AFT and Microsoft Sign Binding AI Privacy Standard for US Schools
The American Federation of Teachers (AFT) and Microsoft have signed a binding agreement establishing ten enforceable privacy and safety standards for AI use in US schools. The deal, announced on September 9, 2026, prohibits Microsoft from using student or teacher data to train AI models and bans tracking students. It requires human oversight for AI decisions and mandates transparency for parents. Microsoft will make these protections available to all US school districts starting November 1, 2026, without requiring contract renegotiation. The AFT describes this as a necessary response to the lack of federal AI regulations.
Key points
- The agreement includes ten legally enforceable principles, with non-compliance resulting in breach of contract liability for Microsoft.
- Microsoft is barred from using 'Covered Data' (prompts, outputs, metadata) to train or improve AI models, with a narrow exception for safety and security.
- The standard prohibits precise geolocation tracking, behavioral profiling, and biometric collection, requiring parental consent for data from children under 13.
- AI systems must have human oversight for decisions affecting students, and features fostering emotional dependency are banned.
- Microsoft must provide plain-language guides for families and ensure data is stored only in the EU or North America.
- The agreement runs for two years and can be renewed in writing; Microsoft must achieve ISO 42001 certification by December 31, 2027.
Background
This development follows AFT's 2025 launch of the National Academy for AI Instruction, which faced initial backlash from some teachers concerned about autonomy. In May 2026, AFT President Randi Weingarten called for screen bans in early elementary schools. The agreement builds on the union's 2024 'Commonsense Guardrails' and responds to recent local actions, including a one-year AI moratorium in New York City schools and a total AI ban in Los Angeles Unified School District.
How outlets are covering it
POLITICO emphasizes the agreement as a 'first of its kind' legal remedy filling the void left by absent federal regulations, highlighting the union's frustration with government inaction. Unite.AI provides a detailed technical breakdown of the ten principles, noting specific compliance requirements like SOC 2 Type II certifications and data storage restrictions. While both sources agree on the binding nature of the deal, Unite.AI notes that Microsoft has not yet completed an independent ISO 42001 assessment for its education AI products, targeting certification by late 2027. The AFT is currently in talks with OpenAI and Anthropic to adopt similar standards, indicating a broader industry shift.
Why it matters
This agreement sets a precedent for private-sector enforcement of AI safety in education, potentially influencing other tech companies and school districts. It provides concrete legal recourse for families and districts, shifting the burden of compliance from vague guidelines to enforceable contract terms. As AI integration in schools accelerates, this standard may become the baseline for national policy discussions.
What to watch
Microsoft plans to make the protections available to all US school districts by November 1, 2026. The AFT continues negotiations with OpenAI and Anthropic to adopt the same standard. Microsoft must achieve ISO 42001 certification by December 31, 2027, and provide plain-language guides for families within 90 days of the agreement's effective date.
- Teachers union reaches AI privacy deal with Microsoft politico.com
- Microsoft, AFT unveil AI safety and privacy standard for schools The Hill
- Microsoft And AFT Put AI Safety Into School Contracts Forbes
- Microsoft Signs Binding AI Safety and Privacy Standard for US Schools unite.ai
- Microsoft, teacher unions unveil national AI safety rules Breakingthenews.net
Want the full story? Read the original reporting
Read on politico.com