The American Federation of Teachers (AFT), United Federation of Teachers (UFT) and Microsoft announced a National AI Safety & Privacy Standard for schools on September 9, 2026. The announced framework would let U.S. school districts incorporate contractual protections into their Microsoft customer agreements, including restrictions on using student and educator data to train AI models, student tracking and AI decisions without human oversight.
AFT President Randi Weingarten, UFT President Michael Mulgrew and Microsoft Vice Chair and President Brad Smith announced the standard. The AFT describes its protections as legally enforceable, but the supplied announcement does not include the full agreement or establish that all U.S. districts have adopted it.
Contents
What the standard would require
According to the AFT announcement, the agreement would establish several categories of protection.
Limits on data use: Student and educator data would not be used to train AI models, sold or repurposed. Model training uses data to build or adjust a model’s learned behaviour. That is different from the operational processing that may occur when an AI service handles information to perform a school task. The announcement clearly addresses training use, but does not provide the agreement’s detailed definitions of training, processing or repurposing.
Restrictions on tracking and automated decisions: The release says students cannot be tracked and that AI decisions require human oversight. Human oversight means that a person remains responsible for reviewing or controlling consequential AI outputs instead of allowing a system to make the decision independently.
The announcement does not specify which forms of tracking are covered, which decisions require review or how much authority a human reviewer must have.
Transparency and control: The standard would require information about how AI is used in schools, what data is collected and what protections apply. Schools would retain control over when and how AI is used, while companies would communicate about product changes, risks and controls.
Security and safety: The listed measures also include security safeguards, protection for sensitive information, human oversight and safeguards against harmful or manipulative experiences.
The release says school districts could end agreements and seek damages from companies that violate the rules.
Why the contract mechanism matters
The announced standard’s main change is not simply a new set of recommended principles. It presents privacy, safety and transparency commitments as terms that districts may add to Microsoft customer agreements.
Contractual safeguards can create obligations between the parties, but their practical scope depends on the exact language, the products and data covered, the remedies available and whether a district adopts the terms. The announcement therefore does not establish a universal protection for every school using Microsoft products.
For a U.S. school district that adopts or negotiates the terms into its Microsoft customer agreement, the standard could make AI governance part of procurement and contract-renewal discussions. Such a school system would have a contractual framework to examine alongside technical features, security controls and educational usefulness.
The approach also gives families and educators a clearer set of issues to ask about: whether their data is used for model training, what information an AI system collects, whether students are tracked, who reviews consequential outputs and how product changes are communicated.
The AFT release frames the agreement as a response to what it characterises as a lack of meaningful federal and state rules governing AI in schools. It also says existing laws including the Health Insurance Portability and Accountability Act (HIPAA) and the Family Educational Rights and Privacy Act (FERPA) did not anticipate the arrival of AI. That is an attributed statement from AFT President Weingarten, not an independent legal analysis contained in the supplied evidence.
What to watch next
The most important next document is the full agreement and accompanying fact sheet. Those materials should clarify the boundaries between data used during normal service operation and data used to train or improve AI models, as well as the scope of the tracking and oversight provisions.
Implementation will depend on which districts adopt the terms, which services are included and what happens when a company or district fails to comply. Until those details and an enforcement record are available, the announcement establishes an announced contractual standard whose nationwide adoption and effectiveness have not been demonstrated.