Frederick County banned student AI use weeks after a Gemini rollout. Here is a practical rollout playbook for districts, and how AI could help with assessment once the groundwork is in place.

Frederick County Banned Student AI Weeks After a Gemini Rollout, and the Districts Watching Are Writing a Better Playbook

A 6 to 1 board vote ended student AI use from pre-K through 12th grade in Frederick County, Maryland, just weeks after a Google Gemini pilot. The story is a useful lesson in sequencing, and assessment experts are already describing where AI could help once districts get the rollout right.

Key Takeaways

•  Frederick County, Maryland's school board voted 6 to 1 this week to ban student AI use from pre-K through 12th grade, with exceptions for lessons about the technology and for accommodations for students with disabilities.

•  The vote came weeks after the district piloted Google Gemini for older students, and board members said parents were not warned about the rollout.

•  Assessment for Good's Temple Lovelace describes AI uses that run behind the scenes, such as generating test questions, scoring faster, and measuring skills like critical thinking.

•  A rollout built on vetting, parent communication, device-level teacher controls, and separate rules for each use case can avoid the whiplash.

A Vote That Followed a Rollout by Weeks

The Frederick County Board of Education in Maryland voted 6 to 1 this week to prohibit student use of artificial intelligence in the classroom from pre-K through 12th grade, according to reporting from Fox 5 DC and TechSpot published on October 9. The ban allows two exceptions: lessons about the technology itself and accommodations for students with disabilities. The board initially considered limiting the ban to younger students before extending it through high school.

What makes the vote notable is the timing. The district had begun piloting Google Gemini for middle and high school students. During this school year, students could use the tool with teacher permission, even though other AI platforms were blocked on school devices. The ban came weeks after that rollout, which turns a policy decision into a case study in sequencing.

What Board Members and Staff Said

The reasons board members gave were specific. They said they were frustrated that Gemini was rolled out without warning to parents. They pointed to AI being used to generate inappropriate images of students and to bully others. One board member described the advice AI gives as bad advice from a mental health standpoint. Teachers, the board noted, cannot disable Gemini on student devices.

The district's director of curriculum and instruction, Scott Murphy, explained that Gemini was chosen because it was the only tool the board evaluated that met the state's safety and privacy standards. That detail matters, because it shows the district did run a vetting process. What the process appears to have missed was communication with families and a way for teachers to control the tool in the classroom.

Educators are already adjusting. Justin Heid, president of the Frederick County Education Association, said teachers are shifting to paper-based work over AI concerns. Many parents support the ban and describe student reliance on AI as a crutch. Others argue that a ban this broad could hurt students' future prospects and leave them less prepared for college. Both views are fair, and both will be argued in board rooms across the country this year.

Frederick County Is Not Alone

TechSpot noted that New York City banned generative AI for elementary and middle school students last month, and that the Los Angeles school district is temporarily blocking generative AI tools on student devices across all grades. We covered the New York decision when it happened, in our look at New York City's pause on student-facing AI, and the pattern is consistent: large districts are choosing to slow down and set rules before expanding access.

That is an encouraging sign for anyone who cares about doing this well. A pause is not a rejection. It is a chance to build the structure that should have come first, and districts that use the pause well will be better prepared when they reopen the question.

A Five-Step Rollout Playbook

The Frederick County story points to a practical sequence that any district can follow. This is our reading of the lessons, not a formula from the district, but each step maps directly to a point of friction in the reporting.

First, vet the tool against written criteria, including your state's safety and privacy standards. Frederick County did this, and it is the right starting point. Second, tell families before launch. A short letter, a page on the district website, and an information night cost little and prevent the surprise that frustrated board members. Third, confirm that teachers can turn the tool on and off at the device level. If a teacher cannot disable it, the classroom has lost a basic control.

Fourth, write separate rules for separate uses. A chatbot a student talks to is a very different thing from an AI system that helps a teacher draft a quiz, and a policy that treats them identically will either block useful work or allow risky use. Fifth, pilot with a review gate. Set a date, define what success and concern look like, and report back to the board before expanding.

Where AI Could Help: The Assessment Conversation

While boards debate student-facing tools, a quieter conversation is unfolding about AI behind the scenes. In an October 8 interview with K-12 Dive, Temple Lovelace, executive director of the Assessment for Good program at the Advanced Education Research and Development Fund, described several ways AI could change assessment. The program focuses on formative assessment for students ages 8 to 13.

Lovelace said generative AI can create test questions that perform as well as those written by subject matter experts, because key source information is used to generate them. She described AI-supported question formats that go beyond fill-in-the-bubble tests, including conversational items where a learner talks through their thinking in a Socratic exchange, which her program has used for about two and a half to three years.

She also said AI can help measure durable skills such as critical thinking, collaboration, and cognitive flexibility, and can score work much sooner than the weeks or months students often wait for summative results. Another idea is a combined learner portrait that brings results from multiple systems together, so superintendents and principals can see individual students in context. Results could then inform what teachers teach next and guide more precise tutoring.

These are one expert's views, and the article did not name specific districts using the tools or offer outcome data beyond Lovelace's claims about item quality and scoring speed. It also did not discuss privacy or bias. That is exactly why the playbook above matters. Assessment uses involve sensitive student data, and the same vetting, communication, and control steps apply.

Why the Distinction Between Uses Is the Whole Story

The Frederick County ban and the assessment ideas can look like opposites, but they sit on the same map. A student chatting with a general-purpose assistant is one use case. A district-controlled system that generates draft questions for teachers to review, or provides faster feedback on formative work, is another. Boards that draw those lines clearly are more likely to protect students and still open the door to tools that save teachers time.

District leaders can use a simple test when evaluating any AI proposal: who is the user, what data does it touch, who can switch it off, and what do families know about it? If those four answers are clear, a pilot is reasonable. If any of them are fuzzy, the decision should wait.

There is also a practical benefit to starting small. A pilot limited to one grade band or one subject gives teachers a chance to report what works, gives IT a chance to test controls, and gives families a chance to ask questions before the program grows.

Who Owns the Decision

AI policy crosses several desks. The chief academic officer and the curriculum director decide how tools connect to instruction. The assessment director weighs in on anything touching testing. The chief technology officer or technology director manages devices, security, and vendor contracts, and instructional technology coordinators support teachers. The superintendent sets direction, and the school board votes, as the 6 to 1 tally in Frederick County shows.

For more on how these roles divide the work, see our guide to the 25 K-12 decision-maker job titles. The practical point is that a vendor, researcher, or partner who wants to be part of this conversation has to reach several of these people, and has to arrive with answers about privacy, controls, and communication.

What Parents Want to Hear Before a Launch

Family communication is the most overlooked piece of a rollout, and the Frederick County reporting suggests it was the most costly gap. Parents do not need a technical briefing. They want to know which tool is being used, which grades and subjects it touches, what student information it collects, what a teacher can do if something goes wrong, and who to call with a question.

A one-page summary that answers those five questions, posted before the pilot begins and sent home in the languages families speak, goes a long way. Districts that add a short information night and a feedback form turn a potential controversy into a record of listening, which helps when the board is asked to expand or continue a program.

Questions Board Members Can Ask Before Voting

Board members do not need to be technologists to ask useful questions. Which written criteria did we use to choose this tool? Can teachers switch it off on a student device? What did we tell families, and when? Which uses are allowed and which are not? What will we measure during the pilot, and when will we report back?

A board that asks those five questions at the start tends to avoid the situation where a vote has to reverse a rollout already underway. It also gives administrators a clear checklist to prepare, which makes meetings shorter and decisions better grounded.

What Vendors and Partners Should Do Next

For companies building AI tools for schools, the lesson is to make the district's job easier. Provide plain-language materials for parents. Offer device-level controls that teachers can use without calling IT. Publish your privacy and safety documentation in a form a board member can read. Be clear about which uses your product supports and which it does not.

The same applies to assessment vendors. If your product generates items or scores student work, expect questions about accuracy, bias, and data handling, and have the answers ready. Districts are listening carefully right now, and credible partners will find that careful buyers are also loyal buyers.

If you want to reach the people who make these decisions, K12 Data can help. Our school district email lists and K-12 database give you education contact data for every role in this story: a chief academic officer email list, an assessment director email list, a curriculum director email list, a chief technology officer email list, a superintendent email list, a school board member email list, a director of instruction email list, and an instructional technology coordinator list. If you are ready to buy school email list data, or want a K-12 marketing database filtered by role, state, and district size, build a targeted list here while districts are still setting their AI policies.

Frequently Asked Questions

Why did Frederick County ban student AI use?

The school board voted 6 to 1 to ban student AI use from pre-K through 12th grade weeks after a Google Gemini pilot. Board members cited a rollout without warning to parents, AI-generated inappropriate images and bullying, concerns about mental health advice, and teachers being unable to disable the tool on student devices.

Can AI be used for student assessment?

Assessment for Good's Temple Lovelace says AI can generate test questions, support conversational question formats, measure skills like critical thinking, and score work faster than traditional summative results. These are one expert's claims, and districts should vet any tool for privacy, bias, and accuracy.

How should districts roll out AI tools?

A practical sequence is to vet the tool against written criteria, tell families before launch, confirm teachers can disable it at the device level, write separate rules for different uses, and pilot with a review date before expanding.