Katy ISD's Age-Gated AI Model: Curriculum Tools Only Until Grade 7

Katy Independent School District, a fast-growing district of roughly 97,000 students in the Houston area, presented a new AI framework to its board on 20 July. It takes effect with the 2026-27 school year, and it does something few district policies have attempted so far: it draws a hard line by age, not just by task.

According to reporting from Houston Public Media and Community Impact, the framework bans open generative chatbots such as Gemini and Copilot outright for students in kindergarten through sixth grade. Younger pupils may use only named, district-approved platforms embedded in the curriculum — Amira Learning and Writable were cited by ABC13 Houston — and only under direct teacher supervision. There is no general-purpose AI access for that age band at all.

Seventh grade is the pivot point. Students there gain access to the same district-approved tools under supervision, but still cannot reach open chatbots on school devices or the school network. Only in grades 8 through 12 can students use generative AI tools for specific educational purposes with teacher authorisation — and Katy ISD requires them to cite when they've used it, turning disclosure into a standing expectation rather than an afterthought.

Why the distinction matters

Most district AI policies published over the past year, including Ohio's statewide mandate and Columbus City Schools' response to it, have left the "what age, what tool" question to individual teacher discretion. Katy ISD's framework instead separates two categories that often get blurred: AI embedded inside a vetted curriculum product, chosen for a specific instructional purpose and monitored by the vendor and the district, versus an open conversational chatbot that can be asked to do or say almost anything.

That distinction has practical consequences. A tool like Amira Learning, built for oral reading assessment, or Writable, built for writing feedback, operates inside guardrails set by its designers and the teachers using it. An open chatbot on a school Chromebook does not. Treating them as different risk categories, rather than lumping all "AI" together under one blanket rule, is a more precise way to write policy than most districts have managed.

A grade-by-grade access model turns "is AI allowed" into "which AI, for what, and who is watching" — a more answerable question for staff enforcing the policy day to day.

Dr Sanée Bell, Katy ISD's assistant superintendent for teaching and learning, told the board the framework ties AI literacy expectations to Texas's Technology Applications TEKS standards, and that a staff AI literacy course is due to launch within weeks of teachers returning from summer break. That sequencing — policy first, then a mandatory training course before the tools reach classrooms — addresses a gap that Microsoft's third annual AI in Education report, published in June, found is widespread: 53 percent of educators globally report they have had no formal AI training, even though 88 percent say they already use AI for school-related work.

The limits worth naming

The framework is new and untested. Enforcement on a network of over 100 campuses is a genuine operational challenge, and "teacher supervision" is only as reliable as staffing and training allow, particularly in a district growing by thousands of pupils a year. Naming specific vendors as approved also creates a procurement dependency: if Amira Learning or Writable changes its data practices or pricing, the district's youngest pupils lose access to any AI-assisted instruction until a replacement is vetted.

There is also no published evidence yet that grade 7 is the right cutoff, as opposed to grade 5 or grade 9. Katy ISD's board presentation cited developmental appropriateness and Texas curriculum alignment, not a study of learning outcomes at different ages. Readers should treat the age thresholds as a policy judgement, not a research finding.

What to do

School leaders drafting or revising AI policy should separate "curriculum-embedded AI" from "open chatbot access" as distinct categories with distinct rules, rather than writing one AI clause that tries to cover both. Pair any access rule with a named, budgeted training course for staff before tools reach classrooms — Katy ISD sequenced training to follow policy by weeks, not months. Where AI is permitted for older students, build disclosure into the assignment brief itself, not into a general honour-code statement, so citation becomes routine rather than exceptional.

What to watch

Whether other large US districts adopt a similar grade-banded model, or whether Katy ISD's approach draws parent pushback from families of older primary pupils who want earlier access. Also worth tracking: whether the named curriculum vendors publish any data on outcomes for the K-6 cohort once the framework has run for a full school year, which would be the first real evidence on whether age-gating changes anything beyond access.

Sources: Houston Public Media, Community Impact, ABC13 Houston, Microsoft AI in Education Report