Digital Learning Program Development

AI Policy


The question of how AI is used in schools is one that schools across the nation are struggling with. New York City famously banned ChatGPT before reversing course and allowing it. Within North Carolina, there is a wide range of policies from Wake County incorporating AI use in the classroom. Other district policies vary widely from allowing the use, to prohibiting it, to suspending students who use it.

The issue of what’s “right” is complex and as we’ve discussed throughout this unit, will vary widely based on the local context. What’s right in Wake County may be impossible in Sampson County and vice versa. With AI in particular, there are many factors to consider when developing a policy such as: local perception from parents and the community, the ability to train teachers to effectively leverage the tools and to teach students to leverage the tools, the instructional goals and needs of the district, readiness to modify lessons to be conducive to inclusion of AI tools, access to technology, the readiness of students to use the tools, other competing priorities, and many other factors.

Schools have very valid concerns about student plagiarism and how students will learn when AI does the work. While many schools have figured out solutions, they’re expensive and take time to implement. There are also open legal questions such as the copyright and plagiarism issues discussed in the previous unit and open questions about FERPA compliance when using AI tools and considerations around vendor Terms of Service with different models.

When considering policies, schools need to consider that AI detectors are highly unreliable exacerbated by some early research suggesting that the more students use AI, the more their original writing style may look like AI. Some professors are using prompt injections to catch students using AI, and others are requiring seated exams. How students use AI, how faculty use AI, and what happens when AI is used for cheating need to be addressed in policy.

Further, ChatGPT and Claude have both created “for teachers” versions of their product that they state are FERPA compliant. While there is no data sharing agreement in place, a teacher accepting the terms of service constitutes a legally binding contract between the school and Claude invoking the “School Official” exception. This also needs to be addressed in policy.

@aarondinin In my mind, the more important question, @Ivy League Roadmap is: if I can’t tell it’s AI, does it matter that it’s written with AI? 🤔 #college #professor #chatgpt #teacherhumor ♬ original sound - Aaron Dinin

Even with teacher use of AI, assumptions may be made by teachers that AI understands how knowledge is developed, the progression of understanding of a topic, or AI-generated lessons may include biases a teacher is unaware of. READ: Education Hazards of Generative AI.

To support districts in generating AI policies, numerous organizations have released guidance. The resources are constantly evolving - use the links below to review each resource:

Using these tools, some school districts have started to generate local AI policies (Example 1, Example 2, Example 3) that create opportunities for students to use AI safely.

In 2026, the NC Budget Bill provided funding for the availability of the MagicSchool AI Platform for all students in grades 6-12, required school districts to adopt a policy based on a sample policy to be provided by the state, and will require all teachers to complete AI training that is being developed for the 2027-2028 school year.