Writing and Updating K–12 AI Use Policies That Meet the Moment

The Status of State-Specific AI Education Policy

Currently, very few states provide guidance on K–12 AI policy. According to MultiState, five states have enacted legislation addressing AI in education, and 22 states and the District of Columbia have introduced bills for consideration.

“New York hasn’t issued statewide guidance, so each district’s board and superintendent decide the guardrails for use applied in their district,” shares Jeffrey Botteron, president and CEO of TechBuffalo, a western New York nonprofit helping build and support the local tech ecosystem. “To make sure decisions about AI use are well informed, we intentionally curate environments for industry and education to collide so the lag between the two shrinks. In Buffalo, we stood up the Future Ready Forum, a monthly meeting to do exactly that.”

LWSD relies on “human-centered” guidance provided by Washington’s Office of Superintendent of Public Instruction to help guide AI decisions.

“Beyond that, the constraints we experience largely predate AI: Student data privacy and data sharing law, records retention requirements, filtering, and consent rules for children under 13 effectively set much of our posture before any AI conversation begins,” says Beebe. “The state guidance gives us ways to think about AI, but the older statutes and rules tell us what we can actually do in this space.”

UP NEXT: See how a Texas school district built a scalable path to AI adoption.

Organizing AI Use Policy in Layers

With little state guidance to reference, most AI policy decisions are left in the hands of district board members and superintendents. To simplify this important but time-consuming process, Miguel Guhlin, director of professional development for the Texas Computer Education Association, suggests a three-layer approach to AI policy: 

1. A foundational layer that emphasizes privacy and safety values
2. An operational layer focused on addressing technology capability changes
3. A classroom layer guiding classroom implementation

“If you have the right foundational policies in place, it doesn't matter how the technology changes. You’ve got your core values to guide you,” Guhlin says. “What does change is how we look at the different AI tools that are out there. Are they worthwhile? Are they good, or not? What are our procedures that we might have around those platforms? These are the guardrails that guide our decisions about AI. Then we need to talk about classroom use of AI. If it doesn't work in the classroom or doesn't give us the benefits we want in the classroom, then we need to have the authority to get rid of it.”

Splitting up AI policy into distinct parts reduces what actually has to be updated. Evergreen district principles and values stay the same or are rarely updated. That leaves only the operational and classroom details that need closer attention on an ongoing basis.

“The evergreen principles should define what AI is used for in your schools and build the uniquely human skills that give anyone who uses AI, teacher or student, a real advantage over someone who uses AI without those skills,” explains Botteron. “The operational layer is the practical part — which tools, which grades and under what conditions. It should change when the technology changes and as teachers get better with it. Review it at least once a year with your teachers and local industry partners, the same way you’d review any curriculum.”

Tapping AI Policy Stakeholders

With AI use policies touching so many wide-ranging topics (including district values, technical details and classroom use), it’s important to include a range of stakeholder perspectives in the mix.

“The working group last year drew mostly from teaching and learning leadership and technology services and technology integration,” shares Beebe. “Teachers and students contributed through focus group conversations, but with less direct representation. Parents and community stakeholders also had perhaps less of a voice at the table, in hindsight, than they should have, so the work this year is to broaden that representation and build on the foundation established last year.”

“TechBuffalo believes that change should be led by the teachers, guided by administration and informed by industry. Teachers can make strong contributions,” says Botteron. “They can tell a simple wrapper on a frontier model from a tool that adds real value. They know which skills will matter and what it takes to coach AI use in a classroom. They belong on the committee that writes your policy and on the one that reviews it each year.”

“Bring in as many different perspectives as you can that represent your school and your community, even if they don't all agree at the outset,” shares Guhlin. “We have to find a way to reconcile these perspectives, and they have to find some kind of expression, especially in that evergreen policy. We need to capture those competing perspectives, test our policy against scenarios and engage in a discussion.”

READ MORE: Educators share how they use artificial intelligence in the classroom.

Evergreen AI Policy Values Drive Operational and Classroom Decisions

Many districts looking for guidance, including LWSD, have turned to the AI Native SAFe Implementation Roadmap to help guide the development of their AI policies.

“The framework itself is sound. The four dimensions are the right questions. What it doesn't provide is a way to decide when those dimensions pull against each other, which is almost always,” Beebe explains. “Safety frequently argues for blocking efficacy for access, and fairness cuts both ways. Meanwhile, the operational questions kept arriving, and we kept answering them case by case, which is exactly what a framework and policy stack are meant to help avoid over the long run.”

It’s these operational questions that drive much of the ongoing need to update AI policy. They continue to surface with each new large language model update. While it’s a challenge to establish and maintain an AI use policy, the more developed and up to date it is, the easier the operational and classroom decision-making becomes.

“We’re seeing more pushback from districts because AI is being embedded into everything,” says Guhlin. “Schools are saying, ‘We didn't agree to have AI embedded in this platform.’ And the districts respond, ‘Well, AI is the future. It's got to be in everything.’ Schools need a way to turn that off. We need to ask ourselves: Operationally, does this align with our values, the evergreen policy that we have? If the answer is no, then we have a hard decision to make. You might have to rule out a vendor if they don't have those shared values or if they can't adapt their product to your district’s values.”