The road to unified AI regulation has been, thus far, a rocky one, to say the least. As I write this article, it is — remarkably — the AI companies themselves pushing for a slowdown in the pace of innovation, while regulators on both sides of the Atlantic are pushing back. I struggle to think of another example of this — at least in recent memory.
That said, despite the disagreements, progress is being made incrementally to ensure that AI serves humanity in as positive a way as possible — as has always been the intention.
One such initiative is the National AI Safety & Privacy Standard, an initiative created by Microsoft, the American Federation of Teachers (AFT), and the United Federation of Teachers (UFT) for schools, students, families, and educators. It’s a great example of how practical, targeted initiatives can help build the broader framework of standards and responsible practices that AI ultimately needs.
National AI Safety & Privacy Standard
News of the new standard was formally announced by AFT President Randi Weingarten, UFT President Michael Mulgrew, and Microsoft Vice Chair and President Brad Smith on September 9, 2026. The importance of the new framework — which we’ll get into shortly — and its emergence at a time of considerable volatility over how AI should be governed at both the national and international levels can perhaps best be summed up by this comment from Weingarten:
“We have forged a hard-fought, iron-clad privacy agreement with real teeth that protects students and families, because no one else, including the federal government, has stepped up to do the real work.
“We can get angrier and angrier, or we can act decisively; anything less than legally enforceable provisions is simply a wish list.”
Described by AFT and Microsoft as a first-of-its-kind agreement, the framework establishes a series of measures designed to protect students, educators, and families.
Among other provisions, the standard prevents student and educator data from being used to train AI models, prohibits the tracking of students, requires human oversight when AI is used to make any high-risk decisions, and requires technology providers to give educators and parents clear, transparent information about how their tools work, what kind of data they collect, and what safeguards are in place.
These new protections are particularly important because of the fact that they are contractually enforceable when incorporated into participating U.S. school districts’ Microsoft agreements. The agreement also gives districts the means to end agreements and to seek damages if the protections are violated.
Let’s be clear, this is not, in itself, a new federal law governing AI in schools, and it doesn’t automatically apply to every technology company. Instead, it’s really a practical model for turning these general principles around AI safety and privacy into concrete, enforceable contractual responsibilities. It’s this kind of incremental progress that could ultimately contribute to a much broader framework for responsible AI.
Final Thoughts
When it comes to determining the trajectory of responsible AI, blanket, universal restrictions won’t work.
It’s far more useful to view the progress of AI regulation as a series of small wins between companies and regulators. These individual areas of focus should gradually build into a broader web of regulations, standards, and codes of conduct that, together, create a complete framework for AI safety and responsible practices.
That, in my humble opinion, is what we should be aiming for.
