A survey of K–12 staff suggests generative AI has become routine office software in many schools before districts have settled on rules for its use. Teachers are using tools including ChatGPT, Google Gemini and Microsoft Copilot to draft lesson plans, adapt materials, prepare special-education supports and write parent communications, according to research led by University of Wisconsin–Stout assistant professor Brett DeJager and reported by StudyFinds.
The work surveyed 303 Wisconsin school professionals and a further 132 respondents across 38 states and Washington, D.C., from spring 2025 through spring 2026. It is not nationally representative, but it provides a useful snapshot of a familiar IT problem: adoption has outrun governance.
ChatGPT was the most commonly reported tool, used by 88% of Wisconsin respondents and 81% of the national group. Respondents also identified Gemini, Copilot, Grammarly and education-focused AI products.
Only 33% of Wisconsin respondents and 29% of the national sample said their district had a formal AI policy. More respondents either reported no policy or said they were unsure whether one existed.
That gap matters because AI use described in the survey goes well beyond harmless brainstorming. Teachers reported generating classroom content and differentiated materials; administrators used AI for emails, memos and training documents; counselors and school psychologists used it for behavioral plans, mental-health resources and individualized education program support.
Those are areas where a polished draft can still contain fabricated facts, inappropriate recommendations, biased language or sensitive information that should not have been submitted to a third-party service. For Windows-centric districts, the presence of Copilot in Microsoft 365 and Windows environments may make that use feel more familiar, but it does not eliminate the need to define approved accounts, data boundaries, retention controls and human review.
The practical issue is not simply detecting AI-written text. DeJager noted in an article for The Conversation that detectors can be unreliable, while a final essay or homework response is no longer dependable proof that a student independently performed the work.
Schools are therefore likely to lean harder on process-based assessment: in-class writing, oral explanations, drafts, source notes and assignments that require students to critique or document AI use. A blanket ban may be easy to announce, but it is hard to enforce when generative features are embedded in consumer apps, search products and productivity suites.
At minimum, districts should publish a short, role-specific policy covering approved AI tools, prohibited data, parent and student communications, required review of generated material, and when students may use AI in coursework. Staff should know whether district-managed Copilot is approved, what Microsoft 365 data it can access, and whether consumer chatbot accounts are off-limits for school information.
The next task for districts is less about choosing a chatbot than giving staff clear rules for using one safely.
The work surveyed 303 Wisconsin school professionals and a further 132 respondents across 38 states and Washington, D.C., from spring 2025 through spring 2026. It is not nationally representative, but it provides a useful snapshot of a familiar IT problem: adoption has outrun governance.
ChatGPT was the most commonly reported tool, used by 88% of Wisconsin respondents and 81% of the national group. Respondents also identified Gemini, Copilot, Grammarly and education-focused AI products.
Policy Has Not Kept Pace
Only 33% of Wisconsin respondents and 29% of the national sample said their district had a formal AI policy. More respondents either reported no policy or said they were unsure whether one existed.That gap matters because AI use described in the survey goes well beyond harmless brainstorming. Teachers reported generating classroom content and differentiated materials; administrators used AI for emails, memos and training documents; counselors and school psychologists used it for behavioral plans, mental-health resources and individualized education program support.
Those are areas where a polished draft can still contain fabricated facts, inappropriate recommendations, biased language or sensitive information that should not have been submitted to a third-party service. For Windows-centric districts, the presence of Copilot in Microsoft 365 and Windows environments may make that use feel more familiar, but it does not eliminate the need to define approved accounts, data boundaries, retention controls and human review.
Cheating Is the Immediate Concern
Academic dishonesty and plagiarism topped respondents’ concerns: 65% in the Wisconsin sample and 74% nationally. Nearly half of Wisconsin respondents and 53% of the national group also flagged difficulty assessing what students actually know when AI can produce finished work on demand.The practical issue is not simply detecting AI-written text. DeJager noted in an article for The Conversation that detectors can be unreliable, while a final essay or homework response is no longer dependable proof that a student independently performed the work.
Schools are therefore likely to lean harder on process-based assessment: in-class writing, oral explanations, drafts, source notes and assignments that require students to critique or document AI use. A blanket ban may be easy to announce, but it is hard to enforce when generative features are embedded in consumer apps, search products and productivity suites.
What District IT Teams Should Do
The survey also found strong demand for free AI training, especially on responsible use, special education, administrative work and assessment. That is an opening for IT and security teams to turn vague restrictions into workable controls.At minimum, districts should publish a short, role-specific policy covering approved AI tools, prohibited data, parent and student communications, required review of generated material, and when students may use AI in coursework. Staff should know whether district-managed Copilot is approved, what Microsoft 365 data it can access, and whether consumer chatbot accounts are off-limits for school information.
The next task for districts is less about choosing a chatbot than giving staff clear rules for using one safely.
References
- Primary source: StudyFinds
Published: 2026-07-19T14:49:43+00:00
Teachers Using AI To Build Lesson Plans, Email Parents
Educators report heavy use of ChatGPT, deep concerns about student learning, and strong demand for free training, study shows.studyfinds.com
