Microsoft agreed with the American Federation of Teachers to adopt new privacy and safety rules for artificial-intelligence products used in schools. The commitments restrict model training, advertising and emotionally dependent chatbot behavior while adding audits and disclosure requirements. The rules are scheduled to take effect November 1 as large school systems pause or review classroom AI deployments.
Microsoft negotiated the commitments with the American Federation of Teachers. The company said student and educator data would not be used to train models except for limited safety purposes. The rules prohibit selling covered data or using it for advertising and unrelated product development. Microsoft agreed to restrict features designed to cultivate emotional dependency among students.
The framework includes audits, transparency reporting and notice requirements. The commitments are scheduled to take effect on November 1. New York City and Los Angeles school systems separately announced yearlong pauses and audits of education AI tools. The agreement applies to Microsoft products and does not automatically bind competing vendors.
School systems often require students to use software, reducing the practical ability to opt out of data collection. Vendor contracts, product defaults and district administration all influence whether a privacy commitment works in practice. Safety exceptions can be legitimate but require clear scope, retention and access controls. A labor-management agreement can establish expectations without replacing federal or state privacy law.
The record remains incomplete in defined ways. The full audit methodology and enforcement remedies had not been tested. The next observable steps are Implementation details before November 1. Whether other major education-technology vendors adopt comparable restrictions.