Microsoft has signed a legally enforceable agreement that places privacy, safety, transparency and human oversight into the contractual relationship between Microsoft and schools. It has entered into this agreement with the American Federation of Teachers (AFT), and that is a significant statement: trust in technology trumps everything else.
In our recent Educators AI Playbook, we heard from Brad Smith, Vice Chair and President of Microsoft, about why this agreement is different and how Microsoft is positioning its technology stack around what is most important to schools, teachers, parents and students. Trust.
In this article, I explore what this means for schools, Microsoft and the wider EdTech market.
Schools have heard plenty of AI commitments over the last few years. Vendors say their tools are safe. They say they respect privacy. They say they support teachers. They say AI will reduce workload and improve learning. Some of those claims may be true, some may be partly true, and some may be difficult to evidence.
The problem for schools is that statements of intent are not the same as contractual obligations, and principles are not the same as enforceable protections.
Microsoft’s move changes the nature of the conversation.
Brad Smith’s core message was that AI is becoming more powerful, more quickly than many expected, and therefore requires stronger action. The agreement is Microsoft’s attempt to set a high bar and go first. It is also a clear positioning statement. Microsoft does not want to be seen as the company telling teachers how to teach. It wants to be seen as the company providing technology that teachers, schools and parents can trust. It clearly sees that as a point of differentiation from its peers.
Much of the AI market is built around the promise that tools can do more of the teacher’s work, provide greater personalisation and save time. Lesson plans, feedback, assessment support, parent communication, pastoral summaries and content generation are all being wrapped in AI.
Some of this will be useful. Some of it may save time. Some of it may support teaching and learning.
But Brad’s point was that schools should never outsource thinking, judgement or decision-making on matters that matter. Microsoft is clearly stating that, from its perspective, AI should support teachers. It should not replace their professional judgement. It should support decision-making. It should not become the decision-maker. In Brad’s words, it should help students learn, but should not weaken their ability to think without it.
The agreement made with the AFT reflects that position. It specifically reinforces that the teacher retains control of the classroom and learning, not AI. The school remains in control of the technology. Students are protected. Parents are informed.
That is a much stronger and more practical position than simply saying AI should be used responsibly.
From 9ine’s perspective, Microsoft’s agreement with the AFT creates three immediate implications.
The first is for Microsoft.
This is a strategic trust move. Microsoft is putting itself forward as the safe, contractual, enterprise-grade AI provider for schools. This is good. Schools need vendors that are prepared to be specific about privacy, safety, data use, security, transparency and accountability.
Microsoft has the scale, legal capacity and technical infrastructure to make commitments such as funding an independent, academically peer-reviewed equity and accessibility analysis on how AI products perform for relevant student populations. Many smaller vendors will struggle to match this and may therefore find it difficult to enter into the same agreement with the AFT.
At another level, this is also highly competitive.
By turning AI safety and privacy into a contractual standard, Microsoft is raising the procurement bar. It is no longer simply competing on product capability. It is competing on assurance.
The message to schools is clear: if you are concerned about AI safety, data training, transparency, retention, deletion, security, human oversight or parent communication, Microsoft can point to a signed standard and ask whether other providers can meet the same test.
That puts the cat among the pigeons.
It also gives Microsoft a stronger answer to the growing technology backlash in schools. Across parts of the US, districts are pausing or restricting student-facing AI. Parents are increasingly concerned about screen time, social media, student data and the effect of technology on learning and wellbeing.
Brad was direct that the technology sector should learn from past mistakes. Social media caused harms that were not always anticipated, but the industry was slow to respond once the problems became clear. Microsoft is trying to show that it will not repeat that mistake with AI.
That does not mean the risk disappears. Microsoft will now need to evidence that the commitments work in practice.
If Microsoft is saying it will not train models on student or educator data, that data minimisation will be built by design, that higher-risk features will be controlled, that families will receive plain-language explanations, and that security will match the stakes, then schools will expect those commitments to be visible, understandable and operational.
The second implication is for schools using Microsoft.
For school districts in the United States, this agreement gives a stronger contractual starting point. It should make it easier to ask better questions, provide reassurance to parents, and evidence that certain AI safeguards are not merely assumed but built into the supplier relationship.
However, it does not remove the school’s own responsibilities.
A school still needs to decide how Microsoft tools are configured, who has access, which age groups can use which features, what data is processed, what is appropriate for students, what staff may use, how parents are informed, what records are kept, and where human review is required.
A contractual standard with Microsoft helps discharge part of the governance burden, but it does not replace local governance.
I believe Microsoft’s approach makes the compliance burden easier in relation to the Microsoft stack. However, at the same time, it reinforces the expectation that schools need to determine how far their other EdTech vendors meet the Microsoft standard.
As such, this may place pressure on those responsible for AI compliance in schools. It is also where the Diamond Formation remains essential.
AI is not one thing.
It is an academic opportunity, a digital safety and safeguarding concern, a privacy issue and a technology implementation question all at once. The academic lead needs to decide whether the tool supports learning, assessment, feedback, workload or inclusion in a meaningful way. The digital safety and safeguarding lead needs to understand whether AI could create harmful interaction, dependency, manipulation, deepfake risk, self-image issues or other digital safety concerns. The privacy lead needs to understand what data is collected, where it goes, whether it trains a model, how long it is retained and what rights apply. The technology lead needs to understand configuration, access, audit, monitoring, security and incident response.
Microsoft’s agreement may provide a stronger supplier baseline, but the school must still decide how AI is used locally. The fact that a vendor meets a high standard does not automatically mean every use case is appropriate.
For schools outside the USA, there is an obvious question: should similar commitments be available in our country?
Much of the Microsoft standard overlaps with the direction of travel already visible through data protection law, cyber security standards, AI product safety guidance and wider governance obligations that have developed in many jurisdictions over recent years.
What Microsoft and the AFT have done is take an area where the USA has no single national AI standard for schools, draw from themes already visible in global privacy and AI governance laws, and turn those themes into contractually enforceable supplier commitments.
That is the significant shift.
The third implication is for the wider EdTech sector.
This agreement may become a new benchmark. Even if schools do not immediately move ecosystems, they will start asking vendors questions that are now standard expectations from the Microsoft agreement.
These questions are not unrealistic. They are exactly the kinds of questions schools should already be asking. What changes is the market expectation.
For schools, the practical next step is not to wait for every vendor to sign something similar. The practical next step is to review your own AI and EdTech estate now. This is the work of AI governance. It is also the reason schools need a cross-functional model such as the Diamond Formation. No single person in a school can answer all of these questions properly. But the right people, working together, can.
Microsoft has placed a marker in the ground. Schools now need to decide whether their own AI governance, vendor assessment and parent communication are strong enough to meet the moment.
You can watch the full Educators AI Playbook discussion on demand here.