On July 21, Redlands Unified School District in Southern California presented a proposed artificial intelligence framework that keeps popular chatbots like ChatGPT and Microsoft Copilot blocked for student use—while laying out a detailed plan to teach AI literacy from kindergarten through 12th grade. The district isn’t ready to hand students the keys to generative AI; instead, it’s building a phased curriculum that treats AI as a subject to understand before it becomes a tool to use.

The Framework: Phased AI Learning Across Grades

Redlands Unified’s plan divides AI education into three stages, each with escalating engagement and responsibility.

  • Grades K–5: Learning About AI
    Younger students won’t touch a chatbot. Lessons will focus on what AI is, where they might encounter it (from voice assistants to YouTube recommendations), and basic safety rules. The goal is awareness, not interaction.

  • Grades 6–8: Learning With AI
    Teachers can use AI-generated examples in class—but only under their own supervision. Students won’t get direct access. A history teacher might share an AI-written essay and ask the class to spot errors, bias, or weak arguments. This turns the tool into a teaching aid, not a shortcut.

  • Grades 9–12: Working Alongside AI
    At this stage, students could begin to use AI directly, with clear boundaries. They’ll learn when it’s appropriate to brainstorm, translate, or revise with AI—and when it undermines real learning. Documentation will be key: students must explain what the AI did, what they changed, and why the final work is theirs.

The model mirrors California’s statewide AI guidance, which recommends building AI literacy from early awareness through critical analysis by high school. But Redlands Unified has added a crucial constraint: the tools themselves remain blocked unless district-approved.

Why ChatGPT and Copilot Are Off-Limits—For Now

Open a browser on a Redlands Unified school computer, and you won’t be able to reach ChatGPT, Google Gemini, or Microsoft Copilot. According to the district’s website, those mainstream chatbots are blocked because they don’t comply with federal student privacy laws—specifically FERPA and COPPA.

FERPA requires schools to protect personally identifiable information in education records, and COPPA mandates verifiable parental consent before collecting data from children under 13. Many consumer AI services, as the district points out, are not designed to meet those standards. Data entered into a free chatbot might be stored, used for training, or exposed in ways that violate student privacy.

Instead, Redlands Unified relies on two vetted tools: Chat for Schools (owned by Skill Struck) and Google Gemini for approved educational uses. The district is also adopting Linewize, a web content filter that will block inappropriate material when the new school year starts. Parents can opt their children out of AI use entirely, and the district can toggle AI access on or off across school properties.

For Windows users, the Copilot block is particularly notable. Copilot is woven deeply into Windows 11, Microsoft Edge, and Office apps—yet for students in this district, it’s off-limits. It’s a reminder that “available” doesn’t equal “school-sanctioned.” Educational IT environments require a layer of control that consumer products rarely offer out of the box.

What This Means for Families, Teachers, and Tech Users

For parents: Your child probably already encounters AI outside school—86% of kids aged 9 to 17 do, yet fewer than half have discussed safe use with an adult, according to research cited by the district and ConnectSafely. A school block doesn’t stop phone use, home searches, or AI hidden inside any random app. The district’s bet is that teaching AI literacy will give students the judgment they need outside the school network.

For teachers: This framework asks educators to move beyond “don’t cheat” warnings. They’ll need training to design assignments that make AI use transparent, create fair grading schemes when some students use AI and others don’t, and handle suspected misuse with evidence—not unreliable AI detectors. The district already advises against using detector products that claim to spot AI-written text, calling them hard to support in a dispute.

For IT administrators and Windows power users: The Copilot block underscores a broader truth: integrating AI into education isn’t about flipping a switch on a consumer feature. It’s about procurement, data governance, account controls, and continuous compliance. Even if a teacher finds a “free” AI tool helpful, it can’t be used until the district vets it. That’s a significant operational lift.

The Push for AI Literacy Before Access

Redlands Unified’s deputy superintendent and chief AI officer, Jason Hill, told the board that the goal isn’t to flood classrooms with AI—it’s “to teach students how to be prepared for a future with it.” That future is already here: 91% of businesses used AI in 2025, Hill noted.

But literacy comes first. A first grader who can’t explain what a chatbot is shouldn’t be using one. A middle schooler should learn to question AI outputs before they’re tempted to copy them. And a high schooler should understand how AI can introduce bias, invent sources, or leak private data—lessons that stick far better than a simple ban.

The California Department of Education backs this sequence. Its guidance urges schools to teach AI as a “subject of study” before making it a “tool for learning,” and to integrate digital citizenship, ethics, and safety throughout.

How the District Plans to Handle AI-Assisted Cheating and Fairness

Board member Melissa Ayala-Quintero asked the practical question: how can a teacher fairly grade one student who used AI and another who didn’t? The answer, Hill acknowledged, will require new assessment models.

One promising element is the district’s plan for five standardized levels of AI use—from “no AI assistance” to “AI as a co-creator.” While details haven’t been released, the idea is that every assignment comes with a clear label. A history paper might allow AI for brainstorming but not for drafting; a coding project might permit AI debugging with detailed comments. This gives students and teachers a shared language, removing ambiguity.

The district will also use a four-stage process for suspected misuse: evidence collection, a preliminary student conference, and an appeal path—deliberately avoiding the trap of treating an AI detector’s score as a verdict. “AI detection tools have proven unreliable,” the district’s own guidance states. Instead, teachers are urged to rely on in-class work, oral explanations, and iterative drafts to gauge understanding.

Board member Candy Olson emphasized the enduring value of pen‑and‑paper check‑ins—a low-tech safeguard that ensures students can still demonstrate knowledge when the screens are off.

Action Steps: What Parents and Educators Should Do Now

While the full policy is still being drafted and won’t be implemented when school opens in August, there are practical moves families and educators can take right now.

For parents:
- Talk to your children about AI—even if the school hasn’t yet. Ask them what tools they’ve heard of, and discuss why a polished answer isn’t always a correct one.
- Review the district’s opt-out process for AI use, and consider whether it aligns with your household’s values.
- If your child uses a Windows PC or Microsoft 365 at home, be aware that Copilot might be available there even if it’s blocked at school. Discuss what’s allowed for homework.

For teachers and school IT staff:
- Review the district’s existing AI guidance on its website. Start thinking about how you might redesign one assignment with clear AI-use boundaries.
- Familiarize yourself with the upcoming Linewize filter and the approved toolset (Chat for Schools, Google Gemini).
- Prepare for professional development: AI literacy isn’t a one‑time training but an ongoing conversation about prompt engineering, bias detection, and data privacy.

Outlook: Training, Transparency, and the Road Ahead

Redlands Unified won’t be ready for a full AI instructional rollout this fall. Teacher training and lesson development must come first, a move that signals restraint rather than delay. The district has acknowledged it still needs to turn the framework into a public, enforceable policy—one that should spell out exactly what data is collected, how long it’s kept, and what happens when an AI vendor changes its terms.

Environmental and financial concerns raised during public comment also deserve a direct response. The district should clarify the cost of its AI tools, the educational problems they solve, and how it will measure success beyond vendor promises.

Ultimately, Redlands Unified is trying to thread a needle that many districts are just beginning to approach: teaching AI literacy without rushing into consumer‑grade tool access. If the final policy matches the framework’s ambition, it could become a model for responsible AI integration—one that prioritizes student understanding over mere availability.