A week after the two largest school systems in the United States pulled generative AI out of their classrooms, the country’s teachers unions moved to set the terms on which it can come back. On Wednesday the American Federation of Teachers and New York’s United Federation of Teachers announced a legally enforceable safety and privacy standard for AI in schools, and Microsoft became the first company to sign it.
The timing is not a coincidence. New York City announced a one-year moratorium on student-facing AI for grades 2-K through 8 on September 2. Los Angeles Unified confirmed the same day that it had cut off generative AI access for all students while it reviews safeguards. The unions’ deal reads as an attempt to answer the question those bans raised: what would an acceptable classroom AI product actually look like?
What Microsoft agreed to
The agreement bars Microsoft from using student or educator data to train its AI models, and from selling that data or using it for advertising. Schools keep control over how data is used, retained and deleted. Microsoft’s education AI tools must carry strong security protections and be designed to avoid harmful or manipulative experiences for students.
The standard takes effect on November 1 and will be available to every district in the country. A district can attach it to a new or existing Microsoft contract without having to renew or renegotiate, which removes the procurement friction that usually slows this kind of change.
AFT president Randi Weingarten called it “a first of its kind” at a Wednesday press conference, and said it exists because federal and state law have not kept up. She wants OpenAI and Anthropic to sign next. Both companies, along with Microsoft, already fund the unions’ National Academy for AI Instruction, a training program launched last year with the goal of making more than 400,000 teachers AI-fluent by 2030.
“You have to have a sense of how to use this, not how it just lands on us and we’re told to do it,” Weingarten said.
What New York actually banned
The New York policy is broader than the headlines suggested. Mayor Zohran Mamdani and Chancellor Kamar Samuels framed it as a moratorium affecting nearly 600,000 students, about two-thirds of the system. All software with student-facing generative AI is out for grades 2-K through 8. Companion chatbots are prohibited in every grade. Screen time is capped at a recommended 30 minutes a day for grades 3 through 5 and 45 minutes for grades 6 through 8.
High school is where the city left a door open. Up to 50,000 general-education students, roughly 5 percent of the system, can take part in five supervised pilots using vetted tools: Quill for close reading, Edia for step-by-step math coaching, Brisk Teaching for teacher-built practice, Playlab for examining AI output for bias, and Intel’s AI-Ready Schools for semester-long projects. Every high schooler also gets two 45-minute AI literacy modules a year covering what the technology is, where it is biased and what it means for careers.
“The tech industry wants us to believe that A.I.-powered early education is not only inevitable, but necessary,” Mamdani said. “We do not see it that way.”
Los Angeles goes further, more quietly
LAUSD’s restriction is total. Before this school year, students aged 13 and up who had completed digital citizenship instruction could use approved generative AI tools. As of the 2026-27 term, none can, on any district device, at any grade. The district’s new Generative AI Ad Hoc Committee held its first meeting on September 2 and is due to hand the board policy recommendations by the end of the school year.
Lila Byock, who founded the parent and teacher group Schools Beyond Screens that pushed for the district’s earlier screen-time limits, said the moratoriums in Los Angeles and New York show that “the people have the power to reclaim US public schools from corporate influence.”
Where this leaves the vendors
Ed tech companies now face a two-sided squeeze. On one side, the two biggest customers in the country have stopped buying student-facing AI for at least a year. On the other, the unions have written a contract clause that any district can adopt, and which the largest vendor has already accepted. A company that refuses to sign the same terms will have to explain why to every superintendent who asks.
Weingarten has been explicit that she supports the bans. In May she called for no screens at all in pre-K through second grade and no AI for any elementary student. On Wednesday she tied the two efforts together, saying the reason New York is pausing children-facing AI is “because we have to get back to how we help develop the muscle of teaching and learning without the cognitive offloading.”
The federal government has so far stayed out of it. In an August letter, the U.S. Department of Education said decisions to limit devices in schools are ultimately local. That leaves cities, unions and vendors to negotiate the rules among themselves, which is exactly what happened this week.
The question for the coming year is whether the November 1 standard becomes a floor that other AI companies match, or a ceiling that only Microsoft, with its existing footprint in schools, could afford to accept. OpenAI’s and Anthropic’s answers will say a lot.
For more coverage of AI in education, visit Mylistingo.
The Netherlands, for internationals
One email a week: the week's news for internationals in the Netherlands, new practical guides and what changed in the cities. No spam, unsubscribe any time.
