Teachers have adopted AI faster than schools have written rules for it. Two Gallup surveys, conducted a year apart, give a reasonable measure of the gap.
The first, run with the Walton Family Foundation in spring 2025, surveyed 2,232 U.S. public school teachers [1]. Sixty percent had used AI for work during the school year, and 30 percent used it at least weekly. Weekly users estimated they saved 5.9 hours a week, which the researchers converted into about six weeks across a school year. That estimate is self-reported rather than observed, so it describes how teachers experience the tools rather than an audit of their time. Even discounted, it helps explain the speed of adoption.
The second survey, conducted in February and March 2026 with 2,069 teachers, asked what guidance those teachers had received [2]. Eighteen percent reported any formal guidance from school administrators on how AI tools should be used. Across ten tasks the researchers asked about, 34 percent received no guidance at all and 48 percent received only informal guidance.
For most teachers using AI, then, the operative policy is something a colleague or an administrator said once in passing.
Why Informal Guidance Causes Problems
Informal guidance creates three problems for schools.
- It is distributed unevenly. A teacher whose department head takes an interest in AI develops a working sense of what is acceptable. A teacher in the next department may develop none. Both are subject to the same expectations, and only one has any basis for meeting them.
- It shifts responsibility to the individual. When guidance is verbal and imprecise, responsibility for a bad outcome falls on whichever teacher used the tool rather than on the institution that never made a decision.
- No one can tell whether the guidance is actually being followed. If a school has never stated what teachers should do with student information in an AI tool, it cannot meaningfully check whether staff are handling that information consistently.
In most systems, AI tools became free and useful before institutions had any position on them, and adoption ran ahead of governance.
What Minimum Guidance Covers
The policy gap has already had a practical consequence: teachers have spent the last two years making these decisions themselves. Which tools are safe enough to use? What student information can go into them? How closely does the output need to be checked? When does a student need to disclose AI assistance?
Schools do not need a comprehensive AI strategy to answer those questions. A few clear decisions would cover most of what teachers are currently missing.
- Write rules for tasks rather than tools. Teachers use AI for lesson planning, worksheet generation, adapting materials, drafting parent communication and assessment design. Those uses are more stable than the products themselves. Schools should decide which tasks are permitted, which require additional review and which are not allowed.
- State what student information can be entered. Can a teacher put a student's name into an AI tool? Their grades? An assessment record or special education documentation? A disciplinary record? Teachers should not have to make those decisions individually.
- Say who is responsible for the output. If AI drafts a report card comment and the comment is wrong, someone is still responsible for sending it. Teachers should remain responsible for material sent under their name.
- Set disclosure rules for both students and teachers. Schools should decide when students must disclose AI use in submitted work, and when teachers should tell students or families that AI was involved in something they received. Most discussion has focused on the first question.
- Keep the prohibited list clear and short. A policy does not need to predict every legitimate use of a new tool. It is more useful to be explicit about what teachers should not do and leave room for uses the school has not yet anticipated.
These decisions will need to be revisited as the tools change. But once a school has made them, teachers no longer have to make the same decisions independently, with no way to know whether their answer matches the school's or the teacher next door.
The Underlying Issue
Written guidance does not undo the fact that teachers have already spent two years figuring out these questions for themselves. It does, however, mean that they do not have to keep answering them alone.
Teachers are already making decisions about AI every day. The question is whether schools are going to leave those decisions to individual judgment or decide which ones should be made collectively.
References
[1] Three in 10 Teachers Use AI Weekly, Saving Six Weeks a Year: https://news.gallup.com/poll/691967/three-teachers-weekly-saving-six-weeks-year.aspx
[2] Most Teachers Receive No Formal Guidance on AI Use: https://news.gallup.com/poll/710534/teachers-receive-no-formal-guidance.aspx
