EdCircuit Staff 
A school AI policy written in 2023 or 2024 may have been entirely reasonable at the time. Many early policies told students to complete their own work, prohibited artificial intelligence unless a teacher granted permission, and classified unauthorized use as plagiarism. Some added a short list of approved uses, including research, translation, data analysis, and grammar assistance.
Those policies responded to the most visible concern confronting schools: students using ChatGPT to write assignments.
Two years later, that concern has not disappeared, but the technology has moved well beyond the problem those rules were designed to solve. AI is now built into search engines, writing programs, productivity platforms, learning management systems, tutoring applications, and accessibility tools. It can generate and interpret text, images, audio, video, computer code, and data. Newer systems can complete multistep tasks, create realistic synthetic media, and interact with students in ways that early policies rarely anticipated.
A district does not need to abandon a policy simply because it is two years old. It does need to determine whether that policy still answers the questions educators, students, and families are asking today.
A Policy Written for Chatbots May Miss Embedded AI
One sign of an aging policy is a broad prohibition on “AI and natural language processing tools” without a clear definition of what that includes. Read literally, the language could cover spell-checkers, automated captions, translation programs, speech-to-text software, search features, and approved accommodations.
It may also overlook the fact that AI is no longer always a separate website a student deliberately visits. A student may encounter an AI-generated summary inside a search result, accept a writing suggestion inside a document, or use a feature added to a platform the district approved before that feature existed.
This creates a difficult enforcement problem. A rule can be so broad that it technically prohibits tools the district expects students to use, yet not specific enough to govern the applications that pose the greatest risks.
Updating a policy does not require schools to permit every new use. It requires them to define what is covered, identify which tools have been reviewed, and explain when AI assistance supports learning and when it replaces the work students are expected to do.
Academic Integrity Now Requires More Than a Plagiarism Rule
Submitting an AI-generated essay as original work can clearly constitute academic dishonesty. That does not mean every unauthorized interaction with AI is identical.
Using AI to generate an entire response, reorganize a paragraph, create practice questions, correct punctuation, or explain a difficult concept involves different levels of assistance. The relevant question is whether the technology replaced the knowledge, reasoning, or skill the assignment was intended to measure.
Districts can give teachers flexibility while providing students with consistent assignment-level expectations. A simple framework might identify AI use as prohibited, limited to specified tasks, permitted with disclosure, or required as part of the assignment. The designation should appear on the assignment so students do not have to guess which rule applies.
This approach also reflects what schools are learning about AI and student performance. A recent review from the U.S. Department of Education’s Institute of Education Sciences found limited but emerging evidence that teacher-mediated AI and tutoring systems that provide guidance or hints may support learning. General-purpose AI may hinder learning when it reduces cognitive effort or completes the thinking students need to practice. The distinction is not merely whether AI was present. It is whether the tool supported the learning process or substituted for it. Institute of Education Sciences
States Are Revising Their Own Guidance
The growth of state-level guidance shows how quickly expectations are changing. In April 2025, the Education Commission of the States reported that at least 28 states had published guidance on AI in K–12 education. By August 2026, AI for Education was tracking official guidance or policies from 35 states and Puerto Rico. Education Commission of the States AI for Education
The more important development is not simply the number of states involved. It is the way some states are revisiting their own work.
California issued initial guidance in 2023, then convened a statewide working group, released updated guidance, and published a model policy in June 2026. The model addresses privacy, disclosure, staff use, approved tools, instructional discretion, AI-assisted grading, and confidential information. It also states that AI detection software should not be the sole basis for discipline or a grade penalty. California Department of Education
Oregon moved its guidance from a static PDF to a web-based Version 3.0 so it could be revised more quickly. The updated material now addresses agentic AI, synthetic media, educator competencies, information literacy, student well-being, bias, privacy, and the risk of becoming overly dependent on automated systems. Oregon Department of Education
Oklahoma has also reached Version 3.0 of its state guidance and supplements it with a model district policy, professional learning, family resources, and implementation support. Its framework connects AI use with FERPA, COPPA, cybersecurity, academic integrity, privacy, and equitable access. Oklahoma State Department of Education
Other states are taking different approaches. Washington has developed human-centered guidance and sample policy information. North Carolina introduced an initial guidebook in 2024 and has continued adding resources and training. Minnesota says its guidance will receive regular updates as the technology evolves, while Colorado describes its AI roadmap as a dynamic resource. Virginia has paired guidance with educator training and system-level leadership support. Ohio requires many public schools to maintain an AI policy and provides a state model they can adapt.
These efforts are not identical, nor are all of them mandatory. Together, however, they send a consistent message: AI governance is becoming an ongoing responsibility rather than a one-time board action.
The Most Important Areas to Revisit
District leaders reviewing an older policy should begin with several practical questions.
Does the policy cover staff as well as students? Educators and administrators may use AI to prepare lessons, draft communications, translate materials, summarize information, provide feedback, or assist with grading. The policy should clarify where human review is required and what information staff may enter into an AI system.
Does it protect student information? Students and staff need to know which platforms have been approved and which information must never be entered into a public tool. Names, grades, education records, health information, disciplinary records, passwords, and unpublished student work require specific safeguards. Vendor review should also examine how prompts and uploaded materials are stored, shared, retained, or used to improve a model.
Does it preserve accessibility and equitable access? A general prohibition should not prevent a student from using technology required by an IEP, Section 504 plan, language-access plan, or other approved accommodation. Districts should also consider whether allowing paid AI products gives some students an advantage unavailable to their classmates.
Does it address high-risk uses? Current rules should cover AI-generated impersonation, deceptive synthetic media, harassment, discriminatory content, unauthorized surveillance, and automated recommendations involving discipline, placement, mental health, special education, or other consequential decisions. Qualified people must remain responsible for decisions affecting students.
Is the enforcement process fair? AI detectors can inform a conversation, but they should not be treated as proof. Schools should consider drafts, version histories, student explanations, demonstrations of understanding, and other evidence before imposing consequences. Broad warnings that students may face civil or criminal liability also deserve review by district counsel rather than being used as a general disciplinary threat.
Keep the Principles Stable and the Procedures Flexible
School boards do not need to rewrite an entire policy every time a company adds a new feature. A more sustainable approach separates lasting principles from frequently changing procedures.
The board policy can establish expectations involving student safety, academic integrity, human oversight, privacy, transparency, equitable access, and responsibility for final decisions. Administrative guidance can maintain the list of approved tools, assignment labels, disclosure methods, vendor-review procedures, data restrictions, and training requirements.
Those procedures should be reviewed at least annually and whenever a major capability, legal requirement, privacy concern, vendor change, or school incident exposes a gap. Students, teachers, technology leaders, curriculum specialists, families, special education professionals, and legal counsel can each identify problems that may not be visible from a single department.
A two-year-old AI policy is not automatically a bad policy. But if it still treats artificial intelligence primarily as a chatbot students might use to write an essay, it no longer reflects the technology operating in today’s schools.
The goal is not to chase every new product. It is to establish a review process that keeps district expectations current while preserving the principles that should not change: students must remain engaged in their own learning, educators must remain responsible for professional decisions, and technology must serve the educational mission rather than quietly redefine it.
Subscribe to edCircuit to stay up to date on all of our shows, podcasts, news, and thought leadership articles.
