For Educators & School Boards · Center for AI Equity and Access

The evidence in two numbers

Stanford’s AI Index asks the world one question: do AI tools bring more benefit than harm? In China, 83 percent say yes. In the United States: 39. And while about a dozen U.S. states have passed laws restricting harmful chatbots — laws we support — the number of states requiring students be taught what this technology is and how to use it well is zero. The United Arab Emirates teaches AI as a required subject from kindergarten. China is writing it into teacher certification.

A blanket ban does not touch the companion chatbots it fears — they were never classroom tools — but it does eliminate literacy instruction, and it strips assistive tools from the disabled students federal law already protects. Excluded once by the condition; excluded again by the policy. We call that the double exclusion.

The Global AI Literacy Gap

The sentiment divide, national mandates abroad, the American asymmetry, and four steps a board can take this year — sources numbered for the packet.

Download (PDF)

Position Paper

Why blanket bans miss their target, the double exclusion defined, and design-based policy explained.

Download (PDF)

In preparationVendor Questions One-PagerFive questions to require in writing

The five design questions, ready to send
What a passing answer looks like
Red flags in vendor language

Facts replace feelings: four steps that make a plan

One principle should govern the decision — the same one boards apply to everything else they oversee. No board sets a reading program, a budget, or a bus route on fear of the unknown; it sets goals, chooses measures, and reviews the evidence. AI deserves the identical discipline. Hesitation is not a plan. This is:

Replace any blanket ban with a design-based policy. Restrict tools by what they are built to do, so the policy reaches harmful systems and spares instructional and assistive ones. Our model policy language is drafted and free.
Adopt an AI literacy scope. Age-appropriate instruction in what AI is, how it produces its outputs, where it fails, and how to use it with integrity. The nations ahead of us did not wait for a mandate.
Invest in teacher capability first. Every national program abroad trains teachers before students. A district’s AI posture is only as good as its faculty’s understanding — and a no-answer tool cannot replace a teacher; it shows the teacher where each student’s thinking stalled, so instruction can finally aim.
Evaluate assistive tools on design merits. Require written vendor attestations and honor existing assistive technology rights in IEP and Section 504 processes.

One policy you can adopt today

No legislation is required for an institution to do what is right. The Center’s Cognitive Access Parity position statement closes with a one-page Model Cognitive Access Policy any school — public, private, or religious — can adopt voluntarily, effective immediately: default access accommodations, a construct-based denial standard, written denials within ten school days, no academic or financial penalty.

Cognitive Access Parity — Position Statement & Model Policy

The principle, the legal asymmetry, the construct standard, and the one-page policy ready for adoption.

Download (PDF)

Five questions to put to any vendor, in writing

Does the tool supply answers? Does it adopt a persona? Does it retain information about the student beyond task progress? What does it do when a student is in distress? Where does student data live? Vendors who built safe tools can answer in one page. The silence of the others is also an answer.

The design tests explained The legislative picture