The Home AI Policy: A One-Page Template for 2026

Three school systems published AI policies in 2026 that are worth stealing from. Washington DC’s Office of the State Superintendent of Education built a red-yellow-green stoplight system that names exactly which decisions AI must never make. Katy ISD in Texas built a grade-tiered access ladder. Illinois published 409 pages that explicitly set no mandates at all. A useful home AI policy template borrows the first two and skips the third, and it fits on one page. Here’s the template, the district language it comes from, and how to fill it in with your kid rather than at them.

Key Takeaways

• DC’s OSSE model policy (September 1, 2026) uses a stoplight: red for high-stakes decisions AI must never make, yellow for limited use with oversight, green for permitted with human review. It’s non-mandatory guidance.
• DC’s red list bars AI from student discipline decisions, teacher performance evaluations, physical surveillance, and determining IEP or Section 504 eligibility.
• Katy ISD (2026-27) tiers access: no generative AI chat K–6, teacher-directed only in grade 7, district-approved tools with teacher permission in grades 8–12.
• Katy also requires students to verify AI output and “disclose or cite AI use when appropriate.”
• Illinois’s 409-page July 2026 guidance “does not lay out rules or mandates,” which is a useful reminder that length is not the same as clarity.

What the districts actually built

DC’s stoplight is the best structural idea to come out of 2026. OSSE released its AI model policy on September 1, 2026 for local education agencies, and it is explicitly guidance, not a mandate: LEAs “are able to tailor the model policy… to adopt it as they see fit.”

The red category prohibits AI for high-stakes decisions that demand human judgment: physical surveillance of students and staff, student discipline decisions, evaluations of teacher performance, and determining eligibility for individualized education programs or Section 504 accommodations. Yellow covers limited use with enhanced oversight: monitoring digital activity on district devices, drafting IEP language, reviewing and grading student work, providing supplemental educator coaching. Green covers permitted use with human review: drafting lesson plans, customizing student-facing materials, developing tutoring plans, analyzing data sets, preparing communications.

OSSE’s framing quote is the one to borrow: the policy “keeps human judgment at the center of decision-making, strengthens privacy and data protections, and empowers educators with guidance they can adapt to their communities.”

Katy ISD’s ladder is the best age-progression model. Presented by Assistant Superintendent Dr. Sanée Bell in July 2026 and effective for 2026-27: in kindergarten through sixth grade, “students will not directly use generative AI chat tools,” though district-approved curriculum-embedded platforms are allowed under direct teacher supervision. In seventh grade, “students begin limited exposure to generative AI, but its use must be structured, teacher-directed and supervised.” In grades 8 through 12, “students may use district-approved generative AI tools for specific educational purposes with prior teacher permission.” Students must follow academic-integrity requirements, verify AI-generated information, and disclose or cite AI use when appropriate. The district’s summary line: “AI can support the learning process, but it cannot replace the thinking, the struggle, or the growth that learning requires.”

Illinois’s 409 pages are the cautionary example. Released July 2026 and developed with a blue-ribbon panel, the document “does not lay out rules or mandates about how schools and districts should use AI,” and its initial drafts were produced using ChatGPT, Claude, and Gemini. State officials disclosed the AI use upfront, which is the right call. But for a parent looking for a rule, 409 pages of non-mandatory guidance is not a rule.

The lesson for a household: one page, with a red list, a ladder, and a disclosure line.

The one-page template

Fill in the blanks with your kid at the table. The bracketed items are decisions; everything else is the structure.

Section	What goes here	Suggested starting content
1. Red list: AI never does this	Decisions and tasks that are always off-limits, no exceptions	Write graded essays or problem sets. Speak for you to a teacher. Decide whether something is safe. Replace a person you were going to talk to.
2. Yellow list: allowed with a check	Permitted with an explicit verification step attached	Summarize a reading you already read (then say what it missed). Check a computation after you solve it. Translate, with disclosure. Generate practice questions from your own notes.
3. Green list: always fine	No approval or check needed	Explain a concept you’re stuck on, in study mode. Quiz you. Read your notes aloud. Help you plan a schedule.
4. The ladder	Which tier this child is on, and what moves them up	[Under 11: adult account only, adult in the room. 11–13: adult account, adult nearby. 14+: own teen account, settings configured.]
5. Disclosure line	The exact sentence to write on work	”I used [tool] for [what], in [which part].” Every time
6. The check	What happens at the end of homework	Two minutes: “walk me through this part.” Every assignment, not just suspicious ones
7. The dose	How often, for schoolwork	[Two sessions per school week, each with a named task from the yellow or green list]
8. Quiet hours	When it’s simply unavailable	[After 9 p.m. on school nights. Platform setting, not just a rule]
9. Recovery path	What happens if a rule breaks	One conversation, one temporary change, one stated route back to normal. Written in advance
10. Review date	When you revisit the whole page	[Ninety days from today. Put it on the calendar now]

Row 1 is the one to spend time on. DC’s insight was that the most important part of an AI policy isn’t the permissions, it’s the short list of decisions a machine must never make. In a school that means discipline and IEP eligibility. In a house it means safety judgments, speaking on your child’s behalf, and replacing a human they’d otherwise have talked to.

Row 5 is the one your kid will resist and the one that pays off fastest. One line on the work turns an ambiguous situation into a documented one, which matters enormously now that Anthropic’s text watermark exists and, per its own documentation, “cannot distinguish ‘Claude wrote this’ from ‘Claude heavily edited this.’”

Row 10 is the one families skip. A ninety-day review is what keeps a policy from becoming a museum piece.

Why the dose is two sessions, and why the check is two minutes

Two numbers in that template come from evidence rather than preference.

The dose. PISA 2025, released September 8, 2026, found that moderate AI users, roughly once a month to once or twice a week for a given task, scored higher in science than both rare users and near-daily users. Near-daily use was reported by about one in five OECD students, and that group scored lowest. Task-level: students who rarely used AI to draft writing scored 509 in science versus 481 for near-daily drafters. The OECD stresses these are associations, not causal proof. Two sessions a week sits inside the band that correlates with the best outcomes.

The check. Roediger and Karpicke (2006, Psychological Science) found repeated testing produced “substantially greater retention than studying,” and that students felt more confident after restudying even though testing worked better. A two-minute oral explanation is retrieval practice, which means the integrity mechanism and the study mechanism are the same activity. The University of Chicago Law School reached the same conclusion institutionally in July 2026, replacing detection software with mandatory oral defenses.

And the reason the policy needs a recovery path rather than escalating consequences: Common Sense Media’s August 2026 data found that when schools blocked AI tools, 44% of teen users hit a block and 59% of those simply switched to a personal device. Prohibition relocates. A defined way back keeps you informed.

Filling it in without it becoming a contract

Let your kid write rows 2, 3, and 7

They’ll write a more permissive yellow list than you would. Let most of it stand. What you’re buying is a policy they’ll actually follow, and what you learn from their draft is more valuable than what you’d have written. Keep row 1 firmly yours.

Say the quiet part about what you can’t see

OpenAI states that linking accounts “does not give a parent or guardian access to a teen’s conversations, chat history, or real-time activity.” Tell your teen you know this. A policy that pretends to omniscience loses credibility the first time it’s tested; one that admits its limits and rests on the disclosure line and the check is more honest and more durable.

Print it and put it where homework happens

Not in a shared doc nobody opens. On paper, on the wall, at the table. Illinois produced 409 pages that nobody will read. One page on a wall gets read fifty times.

Configure the platform settings the same day

A rule on paper plus a setting in the app is much stronger than either alone. For ChatGPT that means Study Mode, Study Hours, Quiet Hours on, and “improve the model for everyone” off. For an Apple device it means a Child Account, Schedules, and Time Allowances. For school accounts it means asking the district what’s enabled.

Put the review date on a calendar, out loud

Ninety days. The review is where a too-strict rule gets loosened and a too-loose one gets tightened, and knowing the review exists is what stops your kid from treating a bad rule as permanent injustice.

What not to do

Don’t make it a contract with signatures and consequences schedules. Katy ISD’s own framing is the better tone: “AI can support the learning process, but it cannot replace the thinking, the struggle, or the growth that learning requires.” That’s a reason, not a threat, and reasons survive better.

What to Watch For Over the Next 3 Months

• Week 4: Is the disclosure line appearing without reminders? That’s the row that indicates the policy is live rather than decorative.
• Month 2 red flags: The two-minute check has quietly stopped, or it only happens when you’re suspicious. Either means the policy has drifted into enforcement mode.
• Month 3 self-check: Do the review. Actually do it. Adjust one row up and one row down, and write the next date on the page.

Frequently Asked Questions

What is the stoplight system in DC’s AI policy?

A three-tier framework in OSSE’s September 1, 2026 model policy: red prohibits AI for high-stakes decisions like student discipline, teacher evaluations, surveillance, and IEP or Section 504 eligibility; yellow allows limited use with enhanced oversight; green permits use with human review.

How do districts handle AI by grade level?

Katy ISD’s 2026-27 framework is the clearest example: no generative AI chat tools in kindergarten through sixth grade, teacher-directed and supervised use only in seventh, and district-approved tools with prior teacher permission in grades 8 through 12.

How much AI use should I allow per week?

PISA 2025 found moderate users, about once a month to once or twice a week for a given task, scored higher in science than both rare and near-daily users. Two sessions a school week, each with a named task, sits inside that band.

Should my child disclose AI use on homework?

Yes, and Katy ISD requires it of its students: verify AI output and “disclose or cite AI use when appropriate.” A one-line note is also protective now that AI watermarks exist and, per Anthropic, cannot distinguish writing from heavy editing.

What if my kid breaks the policy?

Use the recovery path you wrote in advance. Common Sense Media found that when schools blocked AI, 44% of teen users hit a block and 59% of those switched to a personal device, so escalating to prohibition tends to hide the behavior rather than stop it.

Does a written policy actually change anything?

Only if it’s short, visible, jointly written, and reviewed. Illinois published 409 pages of non-mandatory guidance in July 2026, which is a useful reminder that length is not clarity. One page on the wall where homework happens does more.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. OSSE (District of Columbia). (Sept 1, 2026). “OSSE releases AI model policy to guide responsible staff use in schools.” https://osse.dc.gov/release/osse-releases-ai-model-policy-guide-responsible-staff-use-schools
2. Covering Katy News. (Aug 11, 2026). “Katy ISD’s new AI rules: What parents and students need to know.” https://coveringkaty.com/education/katy-isd-s-new-ai-rules-what-parents-and-students-need-to-kn/
3. Capitol News Illinois. (July 15, 2026). “Illinois State Board of Education issues AI guidance, written with help from AI.” https://capitolnewsillinois.com/news/illinois-state-board-of-education-issues-ai-guidance-written-with-help-from-ai/
4. OECD. (Sept 8, 2026). “PISA 2025 Results (Volume I).” https://www.oecd.org/en/publications/pisa-2025-results-volume-i_73451bc5-en.html
5. Fortune. (Sept 11, 2026). “Students are using AI in school and test scores are down.” https://fortune.com/2026/09/11/students-ai-school-test-scores-moderate-use-vs-daily-users/
6. Roediger, H. L., III, & Karpicke, J. D. (2006). “Test-enhanced learning: Taking memory tests improves long-term retention.” Psychological Science, 17(3). https://doi.org/10.1111/j.1467-9280.2006.01693.x
7. Common Sense Media. (Aug 18, 2026). “Teens in the AI Era: Schoolwork and Skills That Matter.” https://www.commonsensemedia.org/research/teens-in-the-ai-era-schoolwork-and-skills-that-matter
8. Anthropic. (Aug 2026). “Claude text watermarking.” https://www.anthropic.com/news/claude-text-watermark
9. University of Chicago Law School. (July 9, 2026). “Rethinking Legal Education in the AI Era.” https://www.law.uchicago.edu/news/ai-strategy-statement
10. OpenAI Help Center. (2026). “Parental controls in ChatGPT: FAQ.” https://help.openai.com/en/articles/12315553

Related reading: DC’s stoplight AI rulebook for schools, Katy ISD’s scaffolded AI access model, and AI rules that survive a personal device.