Most NZ schools are still wrestling with the same question: ban AI or pretend it’s not happening. The trouble is, neither approach quite works. Students are using it anyway. Teachers are doing their best without clear direction. And the Ministry’s guidance, while well-intentioned, can feel like it was written for policy architects rather than classroom teachers.
Here’s what actually works: a traffic light. And a school district in Illinois has already proven it.
🏫 How One School District Solved This
In April 2026, the Consortium for School Networking (CoSN) — alongside ATLIS, ISTE+ASCD, and the Digital Citizenship Coalition — released the third edition of Setting Conditions for Success: Guidelines for Responsible Use of Technology for Schools. It was the first update to address generative AI head-on. But the guidelines themselves aren’t the story.
The story is what Phil Hintz actually built.
Hintz is CTO of Niles Township High School District 219 in Illinois. His team implemented a traffic-light system to communicate AI expectations for every single assignment. Not a school-wide policy. Not a board-level document. A colour, per task, visible to every student before they start working.
“We know students are going to use it anyway,” Hintz said at the 2026 CoSN conference in Chicago. “So we figured, let’s have something in place that actually helps guide them.”
Here’s what it looks like in practice:
- 🔴 Red — No AI allowed. Using it is academic dishonesty, same as cheating. Exam conditions, in-class essays, original analysis.
- 🟡 Yellow — AI is permitted but must be cited. Students share their prompts and AI-generated portions alongside their own work. The AI contribution is visible and marked separately.
- 🟢 Green — AI is required. The assignment is designed so the work is impossible without AI assistance. Students are assessed on how well they direct, evaluate, and iterate on AI output.
The icons appear in the school’s learning management system and on posters in every classroom. Every department — from English to science — has developed assignments at each level. A creative writing piece might be red. A research project might be yellow. A data analysis exercise might be green.
The simplicity is the point. Instead of a single school-wide AI policy that tries to be everything to everyone, the stoplight gives teachers practical, granular control. The framework treats AI not as a monolith to be embraced or rejected, but as a tool whose appropriate use depends on the task at hand.
The CoSN guidelines also push a second shift: from Acceptable Use Policies (AUPs) to Responsible Use Agreements (RUAs). AUPs are board-approved documents in formal, sometimes legal language — top-down, hard to update. RUAs are plain-language, student- and family-friendly, focused on positive behaviours rather than prohibitions. They can be updated quickly as AI capabilities change. The language of “acceptable use” frames technology as a privilege that can be revoked. “Responsible use” frames it as a competency to be developed. For schools navigating AI — where the landscape shifts every few months — the agility matters as much as the framing.
Niles Township isn’t alone. Other districts presenting at CoSN 2026 reported similar implementations. Shad McGaha, CTO of Belton Independent School District in Texas, described how his school uses the stoplight alongside a digital citizenship curriculum that teaches students why the colour matters, not just what it means.
🇳🇿 Why This Works for New Zealand
Three reasons the stoplight fits Aotearoa better than most frameworks:
1. It’s bicultural-compatible. The framework doesn’t assume a single approach. Kura kaupapa Māori can set their own red/yellow/green thresholds aligned with te ao Māori principles — tikanga around original thought, whakapapa of ideas, collective vs individual work. Mainstream schools can do the same with their community values. The colours are universal. The interpretation is local.
2. It’s decile-aware. Here’s what the US experience has shown: students at well-resourced schools are being taught to use AI as a collaborator. Students at under-resourced schools often use it for remediation — or don’t have access at all. The stoplight doesn’t solve this, but it makes it visible. When every assignment has a colour, you can audit whether your school’s green assignments require tools that decile-1 students can’t access at home.
3. It works without waiting for the Ministry. The Ministry’s AI guidance for schools exists but remains high-level and cautious. That’s fine for national direction. But teachers need something they can actually use on Monday morning. The stoplight doesn’t require a board policy change, a consultation process, or a national framework. It requires a teacher to decide, per assignment, what level of AI is appropriate — and communicate it clearly.
📋 Practical Templates for NZ Classrooms
Printable Stoplight Signs
Print these on A4, laminate, stick them on your whiteboard. Students see the colour before they start working.
| Colour | Heading | Subtext |
|---|---|---|
| 🔴 | NO AI | All work must be your own. AI use = academic dishonesty. |
| 🟡 | AI OK — CITE IT | Use AI, but disclose everything. Attach prompts and AI sections. |
| 🟢 | AI REQUIRED | AI is part of the task. You’ll be assessed on how you use it. |
LMS version: Use 🔴🟡🟢 emoji in your assignment titles. Takes 5 seconds per task.
Student AI Disclosure Form
For yellow assignments, require this alongside the submitted work:
AI Disclosure
Name: _______________ Date: _______________ Assignment: _______________
Did you use AI tools? ☐ Yes ☐ No
If yes, which tools? (e.g., ChatGPT, Claude, Gemini, Copilot)
What did you use AI for? (check all that apply) ☐ Finding sources / research ☐ Generating ideas or outlines ☐ Drafting text ☐ Checking grammar or spelling ☐ Writing or debugging code ☐ Translating ☐ Other: _______________
Paste your main prompts below (or attach a screenshot):
What did you change or improve from the AI output?
2-Minute Teacher Checklist
Before assigning work, run through this:
☐ Is the learning outcome about process (yellow/green) or product (red)? ☐ Can students achieve the outcome without AI? (If yes → consider red) ☐ Do all students have equal access to the AI tools the assignment requires? (If no → provide school access or set to yellow) ☐ Does my rubric assess thinking, not just output? ☐ Have I signalled the colour clearly in the LMS and on the board?
⚠️ The Equity Problem We Can’t Ignore
Hintz put it plainly at CoSN 2026: “We’ve been trying to conquer the digital divide of students having access to devices, and we’re doing a pretty good job on that. Now, the new digital divide is that students who can afford the AI are going to have a different set of information than students who cannot afford the AI. That’s going to be an information divide, and that’s really scary.”
He described a direct experience where the free version of an AI model gave him incorrect information, while the paid version provided the correct answer. For students who can’t afford premium AI tools, the quality gap is real and growing.
McGaha flagged a parallel concern: “Students in affluent areas are taught more about how to use AI as a collaborator, while students in lower-resourced areas are using it more for remediation. We must ensure all of our students learn how to steer the AI, not just follow it.”
In NZ, this maps directly onto the decile system. A decile-10 school with enterprise Gemini licenses and 1:1 devices can run green assignments with confidence. A decile-2 school where students share devices and rely on free-tier AI (with its quality limits and usage caps) cannot. The stoplight makes this disparity visible — which is the first step to addressing it.
Māori and Pacific students are disproportionately in lower-decile schools. If those schools only offer red-zone AI policies (bans) while decile-10 schools run green assignments (AI collaboration), we’re not closing the digital divide — we’re deepening it. This isn’t theoretical — as we covered in The AI Price Divide: Who Gets the Best Models?, frontier AI is getting more expensive, not less. The rich get Claude Mythos while everyone else gets last year’s tech. For NZ students, that means the quality gap between free and paid AI tools is widening every quarter.
Some US districts are using purchasing power to bridge the gap — buying enterprise licenses for tools like Google Gemini so every student has access to at least one capable, safe AI environment. But as Hintz acknowledged, this is a patch, not a solution. NZ schools need a systemic answer — and that requires Ministry investment, not just individual school budgets. Subsidised AI access for low-decile schools isn’t a nice-to-have. It’s the difference between a stoplight framework that works for all students and one that only works for the ones who can afford it.
Start Monday
You don’t need a national framework. You don’t need a policy rewrite. You need three colours and a decision per assignment.
Niles Township didn’t wait for Illinois to tell them what to do. They looked at their classrooms, saw students using AI unsupervised, and built something practical. NZ schools can do the same.
Print the signs. Update your LMS. Run the checklist. That’s it. The stoplight isn’t a perfect system. It doesn’t solve assessment integrity or the equity gap on its own. But it gives teachers something the Ministry hasn’t yet: a tool they can use this week, with the students in front of them right now.
And if you’re a school leader reading this — ask yourself how many of your assignments are red, yellow, or green this term. If you can’t answer immediately, your teachers are guessing. And your students are already using AI anyway.
