Three essential early actions

Many education leaders want to get started but are unsure where to begin. Between policy questions, tools, training, instructional shifts, and community concerns, it can feel overwhelming. Arizona leaders built this framework to make the first steps concrete.

Determine and clearly communicate approved AI tools

Starting with a manageable list supports clarity for teachers, students, and families. It also supports safer use and more effective training than is possible when a scattered array of tools is in play. The list will evolve — start it anyway.

Open action one

Train teachers and staff on top-line guardrails

Every AI user needs a few essential understandings: protect data, question outputs, keep human oversight. Treat this as safety training rather than professional development — the deeper work comes later.

Open action two

Clarify expectations for GenAI use

Students must be aware of what the rules are about AI use in school. They notice when expectations vary teacher to teacher, and unclear rules will leave them guessing. Teachers and students should shape these norms together.

Open action three

What "done" looks like

Nine deliverables, drawn straight from the state guidance, grouped under the action each one belongs to. Check them off as your school or district finishes them, then download a summary you can paste into an update.

Approved AI tools, determined and communicated

Done when the list exists, the procedure exists, and both have actually reached the people who need them.

Approved tool list A named, current list of AI tools approved for staff, for students 13 and over, and for students under 13.
Guidance for account creation How accounts get made, with which email, and what staff should do about AI features appearing inside already-approved tools.
Staff, student, and family communication The list and the procedure, actually delivered. A list nobody has seen is not done.

0 of 3 complete

Staff trained on the top-line guardrails

Done when every AI user in the school or district has had the safety conversation, not just the people who came to the optional session.

Intro learning session for all AI users Delivered more than one way, so it actually reaches people who miss the first offering.
Clear do's and don'ts Short enough to remember. Long enough to cover the cases that actually come up.
Data privacy expectations What never goes into an AI tool, stated plainly and tied to FERPA and your own data classification.
Output evaluation basics How staff check AI output for accuracy, bias, and fit before it reaches a student.

0 of 4 complete

Expectations for student use, made clear

Done when a student could tell you the rule for the assignment in front of them.

Transparency expectations When students disclose AI use and how — aligned with your existing academic integrity language.
Communication plan for teachers, students, and families A shared system teachers can use assignment by assignment, explained to students and families before it is enforced.

0 of 2 complete

0 of 9 deliverables complete.

They are the actions that stop being optional

A school or district that skips the tool list ends up with dozens of unvetted apps on its network. One that skips the training finds out that staff are using AI anyway, with student data in the prompt. One that skips student expectations disciplines a student for something nobody ever told them was against the rules.

None of those outcomes require a school or district to have adopted AI. They happen to schools and districts that decided to wait. That is the argument for doing these three first: even a school system that has chosen not to move forward with AI still needs all three.

Doing this alongside other schools and districts is easier.

The Learning Network meets monthly, and it is built for the practitioners actually leading AI implementation in Arizona schools and districts.

Join the Learning Network