> ## Content Index > Fetch the complete content index at: https://akribos.tech/llms.txt > Use this file to discover other available public pages before exploring further. # School AI needs a contract parents can understand - URL: https://akribos.tech/school-ai-privacy-promises-need-receipts/ - Published: 2026-10-30T06:08:36.000Z - Updated: 2026-10-30T12:38:26.000Z - Description: A vendor’s privacy commitment should become something a district can inspect and enforce. - Author: Noah Nelson - Tags: People and technology, Commentary A student should not need to negotiate a technology company’s data practices in order to complete an assignment. When a school selects the tool, the school and the supplier inherit that responsibility. In [September, Microsoft described](https://blogs.microsoft.com/blog/2026/09/16/microsofts-commitment-for-ai-in-education/?ref=akribos.tech) an agreement with the American Federation of Teachers and a Privacy & Safety Standard for Schools covering Microsoft Education products. That is a vendor commitment with a stated scope, not an independent audit of every classroom implementation or a rule binding every AI supplier. The commitment is a starting point. Procurement needs to turn it into evidence and enforceable terms. ## Ask about the ordinary assignment A district should be able to explain where a student’s prompt goes, what gets retained, who can retrieve it, and what happens when the student leaves. Those answers should apply to the actual account and product configuration used in class. Parents need a short explanation that survives contact with ordinary life. Teachers need guidance that distinguishes approved uses from a personal account with different terms. Administrators need a way to confirm that the promised controls are enabled instead of assuming that a brand name guarantees them. An alternative matters too. A student who cannot or should not use a particular service still needs a workable path to the educational objective. That is a design requirement, not a nuisance to address after deployment. ## Translate the promise into purchasing terms Microsoft says its standard limits uses of student and educator data, calls for human oversight of consequential decisions, and provides transparency for families. Those are meaningful commitments to examine. They need to be connected to a named product, the district’s contract, and the configuration actually used in the classroom. A district should be able to point to the terms that govern its deployment without sending a parent through several marketing pages. If a feature invokes another service, the explanation should identify whether the same protections follow the request. Product families are not always a sufficient unit of analysis; the relevant question is which service receives which information under which agreement. The supplier should make changes visible too. A privacy arrangement that was acceptable at purchase can become difficult to evaluate if product names, features, or subcontractors change without a clear record. Schools need a way to review material changes before they become an everyday classroom practice. ## Test learning without borrowing the vendor’s conclusion A fast answer and a successful lesson are different outcomes. Microsoft’s own discussion acknowledges the risk that a tool can remove cognitive work while giving the appearance of learning. That is a useful concern to carry into evaluation rather than treat as a paragraph resolved by the product announcement. A pilot should define the educational task and the evidence of improvement before deployment. A teacher preparing differentiated materials may have a different objective from a student practicing a new concept. Time saved in the first task does not automatically establish learning gains in the second. The evaluation should also count the work created: checking outputs, correcting errors, explaining the tool, and handling students who cannot use it. Otherwise an apparent efficiency gain may simply move labor from one part of the lesson to another. Those costs can be reasonable, but they need to be visible. ## An institutional account is a policy boundary A school-approved workflow can fail when somebody follows a convenient link into a personal account or an unapproved product. The distinction may be obvious to the technology administrator and invisible to the student or teacher. Training should therefore show the actual sign-in path and the signals that identify the intended environment. The U.S. Department of Education’s [guidance on online educational services](https://studentprivacy.ed.gov/sites/default/files/resource%5Fdocument/file/Student%20Privacy%20and%20Online%20Educational%20Services%20%28February%202014%29%5F0.pdf?ref=akribos.tech), published well before the current AI products, recommends evaluating providers and maintaining an understanding of services and information sharing. Its age is a reason not to mistake it for a review of today’s tools. The underlying procurement task, however, remains recognizable. A district cannot evaluate an arrangement it cannot inventory. A manageable list of approved services, their purposes, and their owners is a practical foundation. It also gives teachers a reliable route for proposing a tool instead of improvising around an opaque prohibition. ## The school year eventually ends A product evaluation should include what happens when a class finishes, a student transfers, or a district changes suppliers. The school should know how work is exported, how access ends, and how retained information is handled. A deletion promise should distinguish active records, backups, and any applicable retention requirements. This does not require parents to become contract specialists. It requires the adults choosing the tool to provide an explanation parents can inspect. Students are there to learn. Understanding an invisible chain of commercial data relationships should not become an additional prerequisite for participating. ## Evaluate the lesson and the system A useful AI exercise should have an educational purpose that can be described without repeating the supplier’s product pitch. The privacy arrangement should be just as concrete. Schools already make difficult purchasing decisions with limited staff. Vendors can reduce that burden by making retention, training use, access, and deletion terms easy to compare. A public promise deserves credit when it produces that kind of usable clarity. ## Sources - [Microsoft: AI in education commitment, September 16, 2026](https://blogs.microsoft.com/blog/2026/09/16/microsofts-commitment-for-ai-in-education/?ref=akribos.tech) - [guidance on online educational services](https://studentprivacy.ed.gov/sites/default/files/resource%5Fdocument/file/Student%20Privacy%20and%20Online%20Educational%20Services%20%28February%202014%29%5F0.pdf?ref=akribos.tech)