The American Federation of Teachers (AFT), United Federation of Teachers (UFT), and Microsoft announced on Sept. 9 a national standard designed to give U.S. school districts enforceable artificial intelligence (AI) safety and data privacy protections for students, families, and educators.
Beginning Nov. 1, Microsoft will make the protections available to every U.S. school district. Districts can add the provisions to new or existing Microsoft agreements without renegotiating or renewing their contracts, according to a fact sheet from Microsoft.
The AI Safety & Privacy Standard establishes 10 protections that address AI model training, data collection, data ownership, human oversight, provider accountability, cybersecurity, transparency, accessibility, product changes, and data retention. The agreement allows districts to hold Microsoft contractually accountable for the protections, as well as other AI providers that sign on to the agreement.
Under the standard, student, educator, and customer data generally cannot be used to train or improve AI models. A narrowly defined exception permits certain data processing for safety and security functions, such as detecting self-harm risks, child sexual abuse material, threats of violence, bullying, malicious activity, and security incidents.
“We have forged a hard-fought, iron-clad privacy agreement with real teeth that protects students and families, because no one else, including the federal government, has stepped up to do the real work,” said Randi Weingarten, president of AFT, which represents 1.8 million teachers and other professionals. “We can get angrier and angrier, or we can act decisively; anything less than legally enforceable provisions is simply a wish list.”
The standard also limits how AI providers can collect and retain information. Precise geolocation tracking, behavioral tracking, keystroke logging, long-term profiling, and biometric data collection are prohibited without explicit written approval from the education customer. AI providers must also complete privacy impact assessments before activating material new features that could increase student privacy or safety risks.
Schools retain control over export, retention, and deletion of data covered by the agreement. AI providers must permanently delete requested data from active systems within 180 days, subject to legal holds and other specified exceptions. The standard also requires providers to report data breaches within 72 hours and maintain security controls, including encryption, access controls, independent penetration testing, and incident response plans.
AI systems covered by the agreement cannot make high-risk decisions for students or educators without meaningful human review or prior approval. The standard also prohibits AI companion features and requires providers to give families plain-language information explaining how education-focused AI products work and what privacy, safety, and security safeguards are in place.
“This standard sets a high bar for child privacy and AI safety, and we’ll extend this agreement to every school district across the country,” said Brad Smith, Microsoft vice chair and president.
The agreement applies specifically to generative AI services or features primarily designed and marketed for students, educators, and administrators under an agreement with an education organization. It excludes general-purpose productivity, collaboration, communications, search, cloud, development, and workplace-assistance products that are not primarily designed for education.
The initiative builds on AFT’s 2024 classroom AI guardrails and its National Academy for AI Instruction, which the union launched with Microsoft, OpenAI, Anthropic, and UFT. Earlier this year, Weingarten called for limits on student-facing AI, including a ban in elementary schools and a prohibition on social companion chatbots for students under 16.
“Educators are stepping in to do what we do best: protect our students,” said Michael Mulgrew, president of UFT, the 200,000-member New York City AFT affiliate. “This agreement gives families and school districts the information and privacy protections they have been demanding, and it empowers school districts to end agreements and seek damages from bad actors who break the rules.”
