School AI safety principles move into enforceable contract terms

Read on ⮛

Agreement announced: 9 September 2026 · United States · Reviewed: 14 September 2026

Microsoft, the American Federation of Teachers and the United Federation of Teachers have announced a ten-point AI Safety & Privacy Standard for schools. From 1 November 2026, US school districts can add the protections to new or existing Microsoft agreements and enforce them contractually.

The terms cover limits on using student and educator data for model training, data minimisation, customer ownership of data, human review for high-risk decisions, a ban on AI companions, breach reporting, accessible explanations for families, equity and accessibility, controls against unannounced feature changes, and data portability.

For educators

The agreement has no force in UK schools. Its relevance is as a procurement signal: a major technology supplier has accepted school-specific safeguards as contractual obligations. UK schools and trusts may find the principles useful when asking what data a tool processes, how long it is retained, whether it is used for training, what human oversight exists, how changes are communicated and what happens if the tool becomes unavailable.

Read Microsoft’s announcement.

Read the full AI Safety & Privacy Standard (PDF).

Back to AI in Schools updates

We use cookies to improve your experience. By clicking ‘Accept’, you agree to our use of cookies. Privacy Policy