77 Bills, 27 States: How US Legislators Are Trying to Rewrite AI-in-School Rules

Quick answer

FutureEd's 2026 tracker follows 77 AI-in-classroom bills across 27 US states this session, and two concrete cases show where the rulemaking is landing. Columbus City Schools adopted a formal AI policy in the week of 7 July 2026 - AI as a supplement not a substitute, with teacher discretion per assignment - ahead of Ohio's July 2026 state mandate. Virginia's SB 394 establishes a pilot for practical AI use in public elementary and secondary schools through July 2030. Read together, the 77 bills describe a policy consensus in formation: regulated, taught, disclosed AI use is winning over prohibition, and teacher training, literacy mandates, and procurement rules are the dominant levers.

Key takeaways

• FutureEd's 2026 tracker follows 77 AI-in-classroom bills across 27 US states this session.
• The dominant policy levers are teacher training requirements, mandatory student AI literacy, disclosure standards, data privacy and procurement, and structured pilots.
• Columbus City Schools' policy exemplifies the "supplement not substitute, teacher discretion per assignment" approach ahead of Ohio's July 2026 state mandate.
• Virginia SB 394 chooses the "learn by doing" pilot route through July 2030 rather than prescribing outcomes upfront.
• Very few bills attempt outright bans - the direction of travel is toward regulated, taught, disclosed use.

Why this matters

Two years ago the US state legislative landscape on AI in schools was mostly empty. It now carries a working consensus visible across 77 bills, and two implementation cases already worth studying. That matters because state legislation sets the tempo. When the American teacher-training, literacy-mandate, and procurement templates settle, similar language appears in Australian sector guidance within 18-24 months - as it did with digital-literacy standards a decade ago. Watching what the 77 bills are landing on is the earliest, cheapest signal of where Australian school policy will move next.

For families, the practical read is that a coherent policy floor is being built in real time - not in one jurisdiction, but across many at once - and the shape of it is now legible.

What the 77 bills are actually doing

FutureEd's tracker categorises the bills by focus. The dominant themes:

The shares are drawn from the tracker's public categorisation as of the July 2026 legislative snapshot; the point is the shape, not the exact percentages, which shift as bills advance.

Columbus City Schools: the workable middle

Columbus City Schools' policy, adopted in the week of 7 July 2026 ahead of Ohio's own state mandate later that month, is a clean example of the position most tracked bills are converging on:

• AI is treated as a supplement to teaching and learning, not a substitute for it.
• Teacher discretion per assignment - individual teachers decide whether and how AI may be used on a specific task.
• Applies to teachers, staff and students, so the rules are consistent across who uses the tool.
• Signals to families and staff that the district is not choosing between "ban" and "ignore" - it is choosing "structured, disclosed, teacher-guided use".

The policy is not perfect and does not settle every edge case. What it does is create a workable operating baseline while assessment redesign and teacher training catch up.

Virginia SB 394: the pilot route

Virginia's SB 394 takes a different approach - build evidence before mandating outcomes. The bill establishes a multi-year pilot for practical AI use in public elementary and secondary schools, running through July 2030. In practice that means:

• Selected schools and districts run structured AI-use programmes with reporting requirements.
• The state accumulates case studies, error patterns and cost data that can inform broader statutory choices.
• The pilot's timeline (through 2030) is long enough to see multiple curriculum cycles play out.

Pilots have real drawbacks - they can defer decisions indefinitely - but for a state that does not want to lock in prescriptive rules before the evidence base exists, SB 394 is a defensible instrument. Several of the 77 tracked bills use the same shape.

What this means for schools outside the US

Two implications for Australian and international schools.

First, the policy vocabulary the American cluster is landing on - teacher-training mandate, literacy requirement, disclosure standard, procurement rule, pilot - is the same vocabulary Australian sector authorities will reach for. Getting familiar with the templates now is a preparation shortcut, not academic reading.

Second, the direction of travel matters for admissions. Students applying to US colleges from Australia will meet institutional AI policies that are increasingly harmonised with the state templates their high school system operated under. Students whose Australian schools have already built disclosed-use habits will find the transition simple.

Practical examples

• A US school district in a state whose legislation follows Columbus's template adopts a "supplement not substitute, teacher discretion per assignment" policy and pairs it with a mandated teacher professional-learning cycle - the policy floor and the capability floor arrive together.
• A Virginia public school participating in the SB 394 pilot runs structured AI-integrated coursework in Years 6-8, feeding data back to the state system so future policy is designed against evidence, not intuition.
• An Australian independent school reviews the FutureEd tracker for the specific templates its own board will need to weigh - teacher training, disclosure, procurement - and drafts a local policy that borrows the workable structure.

Common mistakes when reading this legislation cluster

• Counting bills. 77 bills is a headline number; the tracker's real value is showing which levers are dominant, not how many bills exist.
• Assuming state legislation replaces school policy. It sets frameworks; schools still write their own operating rules.
• Reading "pilot" as "no decision". Pilots are decisions - specifically to build evidence before locking in a rule.
• Assuming US legislation won't reach Australia. The vocabulary does, on an 18-24 month lag.
• Ignoring procurement rules. Data-privacy and procurement bills quietly determine which AI tools schools can approve at all - a bigger operational lever than most parents realise.

How the Edison Method applies

Understand. Students learn that policy is a system - state legislation, district policy, school policy, teacher discretion - and that AI rules live at every layer.

Use. Practice happens within realistic disclosure and integrity constraints, so students are already familiar with the operating norms wherever they end up.

Evaluate. Students learn to read a policy document, identify what it does and does not settle, and act accordingly.

Build. Portfolio artefacts include policy-compliant disclosure and process trails, matching what US institutions are converging on.

Lead. Students who understand how policy is being written are the ones best positioned to shape it later - in schools, in workplaces, and in the systems they will help build.

For the deeper US integrity picture, see Brown, UChicago and the AI cheating reckoning. For the Australian policy layer, see AI policy in Australian schools 2026.

The recommendation: don't read the 77 bills as noise. Read them as the earliest visible sketch of the AI-in-education policy environment the whole anglophone school world is heading into.

Sources

1. FutureEd, Legislative Tracker: 2026 State AI in Education Bills, 2026.
2. Inside Higher Ed, Brown professor suspects most of his class used AI to cheat, 8 July 2026 - the tertiary-side pressure.
3. Microsoft Source, Microsoft's new AI in Education report, 24 June 2026 - the training-gap data that is driving many of the bills.