Archive · 15 Sep 2026
Microsoft Sets AI Privacy Rules for Students, Raising Industry Questions
Microsoft made a deal with a large teachers union about how AI should be used in schools.
The deal says student information generally cannot be used to train AI, advertise products or be sold.
It also says schools should not use AI companions that try to make children emotionally dependent on them.
Microsoft must explain its tools clearly to families and allow outside audits.
The rules start November 1 for schools that have Microsoft contracts.
OpenAI and Anthropic say they are working on similar plans.
Google has not said whether it will join.
Some experts support the privacy rules, while others worry that safer AI could still encourage schools to use it before they are ready.
Microsoft agreed with the American Federation of Teachers to adopt privacy and safety standards for AI used in schools.
The agreement bars using student or educator data for AI training, advertising, sales or product development, except for a narrow student-protection exception.
It prohibits AI companions designed to create emotional attachment or dependency and requires audits and plain-language disclosures for families.
The standards apply to schools with Microsoft contracts beginning November 1, while OpenAI and Anthropic are discussing similar agreements.
Google has not said whether it will adopt the framework, as New York City and Los Angeles study or restrict student AI use.
- Who
- Microsoft and the American Federation of Teachers, with discussions involving OpenAI and Anthropic and scrutiny of Google.
- What
- Microsoft agreed to broad privacy, safety, transparency and auditing standards for AI used in schools.
- When
- The standards are scheduled to apply beginning November 1; New York City and Los Angeles also announced yearlong restrictions or pauses as the school year began.
- Where
- The agreement was announced at the United Federation of Teachers headquarters in New York City and applies to schools with Microsoft contracts across the country.
- Why
- The agreement responds to concerns about student data, emotional dependency, transparency and the lack of comprehensive AI legislation for schools.
Supporters of the framework
Critics and non-signatories
Privacy and accountability
Supporters of the framework
Microsoft and the American Federation of Teachers say the legally binding agreement, audits and limits on data use could establish an industry standard and reduce unequal negotiations between school districts and technology companies.
Critics and non-signatories
Critics say the protections will have limited effect unless all major AI companies and the many smaller educational technology providers adopt comparable standards.
Whether safer AI belongs in classrooms
Supporters of the framework
Microsoft says AI should be provided in a form that teachers, students and parents can trust, while the AFT describes the agreement as a major step toward responsible use.
Critics and non-signatories
Fairplay executive director Josh Golin warned that strong privacy protections could be misunderstood as an endorsement of using AI in classrooms, without resolving questions about appropriate ages or whether the tools belong in schools.
Industry participation
Supporters of the framework
OpenAI and Anthropic praised the agreement and said they are working with the AFT on their own safety and privacy pacts.
Critics and non-signatories
Google, a dominant provider of school education technology, has not said whether it will adopt similar protections, despite enabling its Gemini chatbot for some K-12 students using Google Classroom.
- Agreement parties
- Microsoft and the American Federation of Teachers
- Data protections
- Student and educator data cannot be sold, used for advertising or used to develop products; it generally cannot train AI systems.
- Safety restriction
- AI companions or features designed to foster emotional attachment or dependency are prohibited.
- Oversight
- The agreement requires third-party audits and plain-language disclosures to families.
- Start date
- November 1 for schools with Microsoft contracts
- Other companies
- OpenAI and Anthropic are discussing similar agreements; Google has not announced whether it will participate.
- District actions
- New York City and Los Angeles announced yearlong pauses or restrictions while reviewing AI use, privacy and accountability.