A student can ask an AI chatbot to explain algebra, improve an essay or help untangle a difficult homework assignment. In another conversation, that same student might reveal their age, worries, interests or something deeply personal without ever stopping to wonder where those words go next.
That question is becoming harder for schools to ignore.
Microsoft has agreed to sweeping new protections governing how its artificial intelligence tools handle student and educator data, creating what teachers' unions hope will become a new standard for AI in American classrooms.
But Microsoft's promise immediately raises another question: Will the rest of Big Tech make the same one?
Microsoft is putting new limits on what AI can do with student data
The agreement, negotiated with the American Federation of Teachers and the United Federation of Teachers, creates legally enforceable privacy and safety protections for schools using Microsoft products.
Under the new standard, Microsoft says student and educator information will not be used to train AI models, sold for advertising or repurposed for product development, apart from narrow safety-related exceptions. Schools will also retain control over how their data is stored, used and deleted, according to the agreement announced by Microsoft.
The rules go beyond data collection.
AI features designed to encourage emotional dependency or keep children engaged longer than a learning task requires are also prohibited. The agreement requires clearer explanations for families and third-party audits to check whether the protections are being followed.
“This standard sets a high bar for child privacy and AI safety,” Microsoft Vice Chair and President Brad Smith said when announcing the agreement.
Microsoft says the protections will extend to its school district agreements nationwide beginning Nov. 1.
Why teachers wanted something stronger than a promise
Artificial intelligence has moved into education remarkably quickly.
Students are using chatbots to research assignments, brainstorm ideas and explain concepts they do not understand. Schools, meanwhile, are being asked to decide which tools belong in classrooms while the technology itself is still changing.
For teachers' unions, voluntary assurances were no longer enough.
AFT President Randi Weingarten described the agreement as a way of putting control back in the hands of families and educators rather than technology companies.
“We want parents to have control of their kids’ education,” Weingarten said at a news conference announcing the deal. “We want educators to have control.”
The union argues that enforceable contracts matter because school districts should not have to simply trust that a technology company will handle children's information responsibly.
That distinction could become increasingly important as AI systems learn from enormous volumes of data.
Google may now face the biggest question
Microsoft is not the only technology company with a major presence in education.
Google Classroom is used by about 150 million teachers and students worldwide, according to Google figures cited by AP. In August, the company expanded access to its Gemini AI tools for younger students at schools that choose to enable them.
That makes Google's next move particularly significant.
The AFT says it approached Google about the new framework, but as of the latest reporting, the company had not said whether it would adopt comparable contractual protections.
That does not mean Google lacks student privacy safeguards. It means the company has not committed to this particular standard.
Still, once one major technology provider accepts legally enforceable limits, schools may begin asking competitors a straightforward question: Why shouldn't you agree to the same thing?
OpenAI and Anthropic could be next
Microsoft may not remain alone for long.
OpenAI, the company behind ChatGPT, has been discussing its own agreement with the AFT and called Microsoft's move an “important milestone” for AI safety and privacy in schools, AP reported.
Anthropic, which makes Claude, has also said it is working with the union on what it described as a stronger standard for safety and privacy.
If those negotiations produce similar agreements, something bigger could be taking shape.
Instead of waiting for Congress to establish a nationwide AI framework for schools, teachers' unions and technology companies may begin creating an informal industry standard through contracts.
That could put even more pressure on companies that choose not to participate.
AI arrived faster than many of the rules governing it
Federal student privacy protections already exist, but many were created long before generative AI could hold conversations, summarize personal information and respond to children in remarkably human-like ways.
That has opened questions older rules were never specifically written to answer.
Should student conversations ever help train an AI model? How long should those conversations be stored? Can an AI system infer sensitive information even when a student never directly provides it?
Microsoft's agreement offers answers to some of those questions.
What it does not do is settle the national debate.
What Happens Next
The first date to watch is November 1, when Microsoft says the new standards will apply across its U.S. school district contracts.
Attention will then turn to OpenAI and Anthropic as they continue talks with the AFT. Google may attract even greater scrutiny because of how deeply its products are already embedded in schools.
The larger question goes beyond any one company.
AI is finding its way into classrooms faster than lawmakers, educators and parents are reaching consensus about the rules. Microsoft has now drawn one line around children's data.
The next test is whether the rest of the industry decides to stand behind it too.
