In classrooms across America, teachers are firing up ChatGPT to draft lesson plans, write messages to parents, and tailor materials for special education students—yet most school districts still have no official rules governing how these powerful tools should be used. That’s the headline finding from a two-phase study led by University of Wisconsin-Stout assistant professor Brett DeJager, which paints a picture of AI adoption racing far ahead of institutional policy.
The survey, conducted from spring 2025 through spring 2026, gathered responses from 303 educators in Wisconsin and 132 more across 38 states and Washington, D.C. While not nationally representative, the numbers reveal a consistent trend: generative AI is already a routine part of the school day, often with no guardrails in place.
The Silent Infiltration of School AI
ChatGPT was the runaway favorite, used by 88% of Wisconsin respondents and 81% nationally. But it was far from the only tool in educators’ arsenals. Google Gemini, Microsoft Copilot, Grammarly, and a host of education-specific platforms also made regular appearances. In many cases, teachers are using these tools on their own initiative, often through personal accounts.
The activities go well beyond generating quiz questions. Teachers reported using AI to create differentiated materials for learners at various levels and to build resources for students with disabilities. Administrators turned to AI for drafting emails, memos, and training documents. Counselors and school psychologists used it to develop behavioral intervention plans, mental health resources, and components of legally mandated Individualized Education Programs (IEPs).
One respondent summed up the allure: “Love using AI for administrative tasks that take a lot of my mental bandwidth, like parent emails, report card comments, admin tasks, etc.” Another described a more clinical use: “AI helps me breakdown the components of IEP goals so I can create a plan of action for students working towards the goal.”
The speed of adoption is striking. Most respondents had more than eight years of experience in education, yet they embraced AI tools with the enthusiasm of early adopters. This was not a tech-savvy fringe; it was the professional mainstream, finding immediate utility in tools that required no formal training or IT provisioning.
The Policy Void: Only 1 in 3 Districts Have Rules
For all that daily use, formal district policies remain the exception. In Wisconsin, just 33% of respondents said their district had an official AI policy. The national figure was even lower, at 29%. More educators either knew no policy existed or simply weren’t sure—a dangerous ambiguity when the technology is touching student data and classroom instruction.
The gap is especially worrying because many tasks involve sensitive information. Drafted IEPs can contain legally protected details about a child’s disabilities. Behavioral plans may reference private family circumstances. When these are fed into a consumer chatbot like the free version of ChatGPT, there’s no guarantee of data privacy; queries may be used to train future models, and terms of service rarely comply with educational data protection laws like FERPA.
Even for districts that have deployed enterprise versions of AI—such as Microsoft Copilot with commercial data protection in Microsoft 365—the absence of a formal policy means staff may not know what is approved, what remains prohibited, or how to handle generated content that contains errors or biases. Copilot’s deep integration into Windows and Office can make staff feel the tool is inherently safe, but that perception is false without clear governance.
What This Means for You
For teachers and school staff: If you’re using ChatGPT or any consumer AI on your personal phone or laptop to handle school work, you may be inadvertently exposing sensitive student information. Even if the output is useful, the input could violate data-privacy laws and district ethics. It’s critical to ask your IT department which tools are sanctioned and whether your district has a data-sharing agreement with the vendor. Until you get clear guidance, never paste student names, ID numbers, or personal details into a public chatbot.
For IT administrators and technology directors: The survey gives you hard evidence to push for a formal AI policy fast. At minimum, your framework should cover identifying what AI tools are permitted, preventing staff use of consumer accounts for school business, and establishing a human-review requirement for any AI-generated content that will be shared with parents, students, or used in assessments. For Windows-centric environments, you can leverage Microsoft Purview for data loss prevention, conditional access policies to restrict unmanaged devices, and Intune to push configuration profiles that enforce use of enterprise-grade Copilot with data stored within your tenant. Communicate these rules to staff clearly—most say they want guidance but don’t know where to start.
For parents and students: The survey’s top concern—academic dishonesty—shouldn’t be a surprise. When 65–74% of educators list cheating as their top worry, it’s a signal that the traditional take-home essay or homework problem may no longer be a reliable measure of a student’s understanding. Expect to see more in-class writing, oral exams, and assignments that require students to document their use of AI or critique its outputs. As a parent, you can talk to your child about using AI as a tutor, not a crutch, and reinforce that learning remains a human effort.
Why the Explosion Happened So Fast
Generative AI didn’t arrive in schools through a top-down initiative. It came through the side door: free, web-based, and relentlessly promoted by the tech industry. When OpenAI launched ChatGPT in late 2022, educators immediately saw its potential to reduce burnout, a chronic problem worsened by the pandemic. By 2025, Google had embedded Gemini into Workspace for Education, and Microsoft was weaving Copilot into every layer of Windows 11 and Microsoft 365. The tools were suddenly everywhere, and educators, overwhelmed by paperwork, began using them on their own steam.
DeJager’s survey captured the outcome of this organic spread. Teachers and counselors didn’t wait for professional development; they learned by doing. But that bottom-up adoption meant each user made their own decisions about data privacy, pedagogical appropriateness, and ethical boundaries. Some used best practices; others didn’t.
The pandemic-era push to digitize everything also set the stage. By 2025, most districts had already invested in devices and cloud platforms, so integrating an AI chatbot felt like just another app. In a Windows-centric school, Copilot might appear as a sidebar in Edge or a pane in Word, making it feel like just another productivity feature rather than the potentially disruptive force it is.
A Roadmap for Taming the Chaos: Steps to Take Now
The survey found a huge appetite for free AI training—71% in Wisconsin and 63% nationally said they wanted it. That demand is an untapped lever for change. IT leaders can use it to move beyond vague restrictions and toward workable controls.
Immediately, districts should publish a short, role-specific AI policy that answers:
- Which AI tools are approved for school work? (e.g., district-licensed Copilot, not free ChatGPT)
- What data must never be pasted into an AI tool? (student PII, health records, confidential personnel info)
- When must a human review and edit AI-generated content before sharing with parents or students?
- Under what conditions may students use AI in coursework?
For technology staff, the next steps are concrete:
- Audit current usage: A quick survey of staff can reveal how many are using ChatGPT or other tools with their school accounts. This isn’t to punish but to understand the gap.
- Configure Microsoft 365 settings: If your district pays for Copilot for Microsoft 365, enable commercial data protection to ensure chat data isn’t used to train foundation models. Use Azure AD conditional access to block sign-ins from personal Microsoft accounts on managed devices.
- Block consumer AI sites (with caution): A blanket network block on ChatGPT.com may drive users to mobile hotspots, but it signals that the tool isn’t sanctioned. Combine this with a sanctioned alternative.
- Roll out a training module: Microsoft Learn and other platforms offer free modules on responsible AI. Make one required for all staff, covering data privacy, bias, and citation practices.
- Involve curriculum leaders: Work with instructional coaches to develop model assignments that teach students how to use AI ethically, rather than pretending it doesn’t exist.
For teachers, while you wait for official policy, you can adopt your own “safe AI” habits: use only district-provided accounts, run all generated content through a fact-check, and keep a log of AI use for transparency. If your district does nothing, advocate for a policy through your union or department head.
Looking Ahead: AI Is Not Leaving the Classroom
The survey’s timeline ends in spring 2026, but the trend has only accelerated. Microsoft has signaled even deeper Copilot integration in upcoming Windows 11 updates, and Google is pushing Gemini across Chromebooks. District policies will inevitably catch up, but the delay is costly.
Schools that craft and communicate clear policies this year will reduce their legal liability, protect student privacy, and ease teachers’ anxiety. Those that drag their heels risk data breaches, grade-integrity scandals, and a widening gap between what students do at school and what they can do at home with AI.
The appetite for training is loud and clear. Districts have the chance to turn AI from a free-for-all into a managed tool that genuinely boosts efficiency without sacrificing ethics. The survey’s most important finding isn’t the 88% using ChatGPT; it’s that educators themselves are calling for the rules to catch up.