A North Dakota teacher joins Mark Cuban's AI fellowship, amid the legal vacuum in U.S. classrooms

By Zendoric · September 14, 2026.

Kristy Sandbeck has been teaching for more than 30 years at Grand Forks Public Schools (GFPS), the school district of this North Dakota city. Last week she was selected for the Mark Cuban Foundation Teacher Fellowship, a national program that brings together teachers from every subject to assess and define best practices on the use of artificial intelligence (AI) in the classroom, the Grand Forks Herald reports. Sandbeck, a mentor at the district's Virtual High School, thus joins "dozens" of teachers from across the country taking part in the initiative.

The fellowship is the work of Mark Cuban, the investor and "Shark Tank" host who in recent years has redirected his philanthropic foundation toward AI. In 2019 Cuban launched a series of bootcamps to teach high school students to use the technology responsibly; in 2025 he expanded the program with this fellowship for teachers. Sandbeck will run one of those bootcamps in Grand Forks next year, with no date set yet.

In practice, this translates into very concrete uses. Sandbeck cites students who have used AI to learn how to take apart a car engine, build models of the solar system or turn a text into audio or video to reinforce reading comprehension. For teachers, the technology serves to draft lesson plans, grading rubrics or individualized education programs (IEPs), the documents that tailor instruction to special education students. Her message, however, is one of caution: "we can't forget human intelligence, that's a very important thing to understand," she told the Herald; "AI is there, but we can't depend on it all the time."

That caution has statistical backing. According to a 2025 study by the College Board —the nonprofit that administers the SAT and Advanced Placement exams in the U.S.—, 84% of high school students already use generative AI tools to do their homework. The same study found that fewer than half of the country's high schools have an official policy governing that use. GFPS does have one: it allows models such as ChatGPT (OpenAI) or Claude (Anthropic) only with the teacher's authorization, and the teacher must also spell out in the lesson plan how the tool will be used.

That, in fact, is the underlying story. Sandbeck's fellowship is not a major product announcement or a funding round: it is a symptom that AI adoption in classrooms has run well ahead of the rules meant to govern it, and that the gap is being filled, for now, by private philanthropy rather than public policy. That it is a billionaire's foundation, and not the Department of Education or individual states, organizing bootcamps and funding fellowships to "teach how to teach" with AI says something about the pace at which each actor moves: fast on technology, slow on administration.

In the short term that mismatch carries real costs that should not be minimized. Usage of 84% with no clear framework in more than half of schools opens the door to academic integrity problems that are hard to audit. It also creates inequality between districts: those that, like GFPS, already have their own policy and now a teacher trained through a national fellowship, and those that have neither. For a while, the quality of AI-enabled education a student receives will depend more on the luck of the district they live in than on any common standard.

In the long run, however, the examples Sandbeck gives point to something more promising than a mere classroom-management patch. Adapting a text to each student's reading level, turning it into audio or video, or quickly generating an individualized education program are, in essence, forms of personalization that until now only those paying for a private tutor could afford. If that capability becomes widespread and is well governed, AI does not replace the teacher: it frees up time from administrative work and gives it back to what Sandbeck calls "human intelligence," the judgment and support no model replicates. It is the same divide we have been observing in other sectors —those who use AI as support come out ahead, work that consisted only of repetition disappears— and in education that transition is still being written district by district. This week it was Grand Forks's turn.