🔍 Read the full analysis: The Human Control Rules In Microsoft’s School Pact And California’s New Laws on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Microsoft says that from Nov. 1, 2026, every U.S. school district can use a contract standard restricting how education-focused AI handles student data and makes decisions. The agreement is contractual, not a law, and its scope excludes general-purpose productivity and workplace tools. Separate California and New York laws set rules for AI in employment, health care and services used by children.
Microsoft says U.S. school districts can use a new AI contract standard starting Nov. 1, with terms barring covered education products from training on student data, tracking pupils or making consequential decisions without meaningful human review. The agreement is a contract framework, not a new law; it arrives alongside separate California and New York laws addressing AI in workplaces, health care and child-facing chatbots.
The 31-page Memorandum of Agreement was announced Sept. 9 by the American Federation of Teachers, its affiliated United Federation of Teachers, the National Academy for AI Instruction and Microsoft. The source account describes the Academy as an educator-led training body announced by the AFT in 2025. Microsoft says it is making the terms available to every U.S. school district from Nov. 1. Districts can incorporate them into their own Microsoft contracts. The document also has empty signature slots for other AI providers, but the supplied material does not say that any other provider has signed.
For products covered by the agreement, student, teacher and administrator prompts, responses, files, behavioral signals and metadata may not be used to train, fine-tune, update, benchmark or otherwise improve an AI model. The restriction continues after a contract ends, and the text says de-identification is not an exception. A narrow safety provision allows processing to identify matters such as self-harm risk, abuse material, grooming, threats, bullying or security attacks; the information cannot then be used for general model improvement. The agreement also bars specified forms of location and behavioral tracking, attention monitoring, long-term profiling and biometric or emotional inference, subject to an exception requiring a school customer’s explicit written approval.
The terms prohibit companion-style features designed to foster emotional attachment or dependency and require meaningful human review or prior approval for decisions about students or teachers. Student-facing agents that act outside a provider’s systems, such as by sending messages or making purchases, must be off by default unless an administrator enables them. The agreement sets a 72-hour breach-reporting deadline and requires deleted data to be removed from active systems within 180 days. It also specifies data-location limits covering the EU or North America. The source says the Academy will instruct educators not to use AI for automated grading without review, discipline, placement decisions or emotional assessment.
Infographic 01 · The lines drawn this autumn
Seven weeks, one direction: a human stays in charge
From the school contract to state laws: what was signed when, and what the classroom standard actually requires.
- 2 Feb 2025EU: emotion inference banned at work and schoolAI Act Art. 5(1)(f), except medical or safety reasons
- 9 Sep 2026AFT, UFT and Microsoft announce the school standardA contract districts can add to their Microsoft agreements
- 30 Sep 2026California signs AB 1883, SB 947, SB 951, AB 1979Emotion AI at work, robot bosses, AI layoff notices, AI in clinics
- 28 Oct 2026New York signs its chatbot law for minorsS.9051B/A.10379C: no unsafe companion features for under-18s
- 1 Nov 2026Microsoft makes the standard available to all US districtsPer Microsoft’s fact sheet; districts “can add them”
- 1 Jan 2027Most California rules take effectAB 1883, AB 1979, SB 951
- 1 Jul 2027SB 947 becomes operativeHuman corroboration before AI-driven discipline or firing
What Microsoft’s classroom AI may not do
- Learn from students or teachers. No training on their data; de-identification is “not a loophole”.
- Track them. No keystroke logging, attention monitoring, long-term profiling or biometrics.
- Befriend them. No companion features designed to foster emotional attachment.
- Decide alone. No decisions without meaningful human review or prior approval.
Where it is narrower
- Scope: education products only. General-purpose productivity and workplace tools are excluded by definition.
- Opt-in or automatic: Microsoft says districts “can add” the terms; the agreement says “upon request within 90 days”.
- Exceptions: a narrow safety exception for training; the no-tracking list allows written approval by the school.
“Anything less than legally enforceable provisions is simply a wish list.”Randi Weingarten, AFT President, 9 September 2026
Sources: AFT/UFT/Microsoft release, 9 Sep 2026 · Standard MOA v1.0 · Microsoft fact sheet · Governor of California, 30 Sep 2026 · Governor of New York, 28 Oct 2026 · EU AI Act, Art. 5(1)(f). Effective dates for California follow the chaptered texts and the state’s default start date for regular-session bills.
Powered by Thorsten Meyer AI
Contract Terms Put Districts in Control
The standard gives school districts a way to attach specific safeguards to procurement agreements rather than relying only on a provider’s general privacy statements. If a district incorporates the provisions into its contract, the source says it may be able to end the agreement and seek damages for a breach. That makes the protections potentially enforceable through the contract, but it does not make them a nationwide legal requirement for every school or every AI product.
For families and educators, the provisions address several high-stakes questions: whether student work can feed model development, whether software can monitor behavior or infer emotions, and whether an algorithm can make decisions affecting a child’s education without a person reviewing them. They also set limits on systems designed to simulate friendship, a concern distinct from accuracy or data security.
The employment and health-care laws described in the source extend the same human-oversight debate into other settings. California’s restrictions could affect employers using automated tools, while its health-care provision draws a boundary around tasks reserved for licensed professionals. New York’s child-focused chatbot law addresses relationship-like features. These are separate measures with different legal mechanisms, coverage and dates; they should not be treated as one national AI code.
Products Worth Considering
BowFlex Results Series SelectTech Adjustable Dumbbells (Pair)
Oral-B iO Series 3 Rechargeable Electric Toothbrush for Adults, Black
Levoit Vital 100S-P Air Purifier with Washable Pre-Filter
As an affiliate, we earn on qualifying purchases.
Three Measures, Different Legal Reach
The school agreement is a memorandum between the National Academy for AI Instruction and Microsoft, rather than legislation or a general Microsoft policy page. Its coverage is limited to AI products designed and marketed for education. The source explicitly excludes general-purpose productivity and workplace tools, so the terms should not be assumed to cover every Microsoft product used by a student or employee.
The source says there is a difference between the agreement’s description, Microsoft’s account and press reports over whether the protections apply automatically or only when a district requests them. It does establish that districts can put the terms into their Microsoft contracts. The practical force of the standard therefore depends on contract adoption and the products covered.
California Gov. Gavin Newsom signed four measures on Sept. 30, according to the supplied material. AB 1883 restricts AI emotion-reading of workers and collection of neural data; SB 947 bars discipline or firing decided by an algorithm alone; AB 1979 limits AI from performing health-care work reserved by law for licensed professionals; and SB 951 requires layoff notices to disclose when AI replaced jobs. Most are scheduled to take effect Jan. 1, 2027, while SB 947 is scheduled for July 1, 2027. New York Gov. Kathy Hochul signed a separate law on Oct. 28 targeting what the source describes as unsafe companion features for minors and unverified users.
““Anything less than legally enforceable provisions is simply a wish list.””
— AFT President Randi Weingarten
Adoption and Coverage Still Matter
It is not clear how many districts will adopt the terms, whether Microsoft will apply them by default or only when a district requests them, or whether other AI providers will sign the agreement. The source itself says accounts differ on the default-versus-request question. Until districts make procurement decisions and contract language is available for specific deployments, the standard’s reach cannot be measured.
The text also allows school customers to approve exceptions in writing to some prohibited tracking practices. The source does not explain how districts will use that discretion or what safeguards they would require. It does not provide evidence about compliance monitoring, disputes, or enforcement outcomes. For the state measures, the source gives signing dates and scheduled effective dates but not implementation guidance, court challenges or agency rules. Those details may affect how the laws operate in practice.
District Procurement and State Rollout
From Nov. 1, districts can decide whether to incorporate Microsoft’s terms into their contracts and which education products the provisions will cover. Schools and providers will need to resolve the practical questions around data handling, safety exceptions, human review, breach notices and deletion deadlines in each covered deployment. The supplied material does not name a deadline for districts to act or a review process for checking compliance.
California’s measures are scheduled to take effect in stages, beginning Jan. 1, 2027, with SB 947 following on July 1. Employers, health-care organizations and technology vendors will need to track the requirements applicable to their uses of AI. New York’s signed law will also require further attention to how its restrictions on child-facing companion features are interpreted and applied. For now, the confirmed development is a new contract option for schools alongside state-level rules—not a single, uniform set of AI requirements.
Key Questions
Is Microsoft’s school AI standard a law?
No. It is a 31-page contractual agreement between the National Academy for AI Instruction and Microsoft. A district can incorporate its terms into a Microsoft contract; the agreement does not itself create a nationwide legal mandate.
What does the agreement prohibit for covered education AI?
It bars using student and educator data to improve AI models, restricts specified tracking and emotional inference, prohibits companion-style features designed to foster dependency, and requires meaningful human review or prior approval for decisions about students or teachers. It includes a narrow safety-processing exception and a written-approval exception for some tracking restrictions.
Does the standard cover every Microsoft tool used at school?
No. The supplied account says it covers AI products designed and marketed for education. General-purpose productivity and workplace tools are expressly excluded, so coverage depends on the product and the district’s contract.
What do the California laws address?
The four measures cover worker emotion and neural-data collection, algorithm-only firing or discipline, AI performing health-care work reserved for licensed professionals, and layoff notices that disclose when AI replaced jobs. Most are scheduled to take effect Jan. 1, 2027; SB 947 is scheduled for July 1, 2027.
Have other AI companies joined the school agreement?
The source describes empty signature slots for other providers but does not identify any that have signed. Whether the standard will extend beyond Microsoft’s products remains unclear.
Source: ThorstenMeyerAI.com
