Schools Need Boundaries Before Wider Adoption

New teacher tools, student policy proposals, district rules, and classroom studies all arrived this week. They showed that access is expanding before schools have agreed on the decisions AI should never make.

What Happened This Week

On Tuesday, July 14, Anthropic announced Claude for Teachers, a free, standards-aligned version of Claude for verified K through 12 educators in the United States.¹²

Last week in New York City, at the Bridges public sector AI summit, a student sat on a panel of district leaders and described what it felt like to be handed a seat at the policy table.³⁴

On Monday, July 28, the Hechinger Report's Proof Points column published a piece with a headline that will not have been read by most district leaders and should be.⁵⁶

On July 20, Katy Independent School District, a Texas district of ninety thousand students west of Houston, released an artificial intelligence framework for the 2026-27 school year that does something almost no other American district has done this summer.⁷⁸

On July 21, at InstructureCon in Louisville, the learning management company Instructure released a survey that put two numbers next to each other.⁹¹⁰

On July 28, three days before the fall semester was scheduled to start, the Orange County Public Schools board in Florida approved its first AI policy for the coming year.¹¹¹²

On July 28, The Hechinger Report covered the first randomized field trial of a generative AI teaching assistant in real classrooms.⁵⁶

What These Stories Have in Common

AI can support planning, tutoring, and feedback, but it should not quietly take over grading, discipline, or judgments about a student's ability. Those decisions require context and accountability that a model does not have.

Koan starts from a simple premise. The final answer is no longer a reliable measure of learning. Schools need to understand how students reached that answer and whether the work made them more capable.

What Schools Can Do

Before expanding access, schools should define clear boundaries for high-stakes decisions and make every AI-supported action reviewable. They should also test whether the tool improves student learning rather than only teacher efficiency.

A school should decide what remains human before a product decides it by default.