52 Bills, Zero Communication Theory: The Legislative Capture of AI Literacy

Twenty-five U.S. states have introduced 52 bills related to AI in education during the 2026 legislative session. Every one of them defines AI literacy as a form of tool proficiency. Not a single one applies communication theory. The result isn't education policy — it's vendor certification with legislative backing.

The Numbers

FutureEd's legislative tracker documents the surge: 52 AI-in-education bills across 25 states in 2026 alone. The bills cover teacher training, student competencies, curriculum mandates, and institutional AI policies. Some ban AI in classrooms. Others require it. What they share is a definition of "AI literacy" that reduces to: can the student use the tool effectively?

That sounds reasonable until you ask who designed the tools, who wrote the curricula, and who defined "effectively." The answer to all three is increasingly the same companies.

The Vendor-to-Legislature Pipeline

Here's the cycle: An AI company builds a product. The company (or a company-funded nonprofit) develops "AI literacy" curricula teaching students to use that product category effectively. State legislators adopt the curricula frameworks into law. Schools implement what the law mandates. Teachers attend vendor-aligned professional development. Students learn to be fluent consumers of specific products.

At no point in this pipeline does anyone ask: What does it mean to communicate through an application layer? The question doesn't arise because the pipeline isn't designed to produce communication. It's designed to produce adoption.

EdWeek survey data from 2026 confirms the trend: AI literacy is "becoming the norm" in K-12. But "norm" means institutionalized — and institutionalization without theory means the current practice gets frozen into policy. Whatever Google, Microsoft, and OpenAI happen to mean by "AI literate" is what legislators will codify.

Training ≠ Education

The distinction matters. Training teaches you to operate within a system's design assumptions. Education teaches you to interrogate those assumptions. Training produces fluent users. Education produces fluent thinkers.

When the vendor defines the curriculum, "literacy" means "comfort with our product category." That's not literacy in any pedagogical sense. It's customer onboarding dressed in academic language.

Holly Clark, keynoting Spring CUE 2026 on AI literacy, puts it well: "Bans don't build judgment." But the alternative she and others advocate — structured AI integration — still centers on tool use. The hidden equity gap isn't between students who use AI and students who don't. It's between students who learn to navigate AI systems and students who learn to consume them.

What Communication Theory Would Add

Application Layer Communication treats every software interaction as a communicative act. You're not just "using ChatGPT" — you're navigating an environment shaped by training data, RLHF, system prompts, and API constraints. The quality of that navigation depends on communicative fluency: Can you read the system's affordances? Can you adapt when it misunderstands? Can you recognize when the interface is shaping your request before you make it?

None of the 52 bills address this. They address what students should do with AI tools, not how the communicative relationship between student and tool works. The omission isn't accidental. Communication theory would require examining the power asymmetry embedded in tool design — and the vendors writing the curricula have no incentive to teach students to see it.

The Stratification Mechanism

This is where it gets structural. Students from well-resourced schools with technically fluent teachers will develop informal communicative literacy through practice — they'll learn to navigate, not just use. Students from under-resourced schools will get the legislatively mandated minimum: tool proficiency. Same tools, radically different relationships to those tools.

Hancock et al. (2020) demonstrated this dynamic in AI-Mediated Communication: the same AI system mediates differently based on who's navigating it. Cotter & Reisdorf (2020) found that breadth of technology use predicts algorithmic knowledge five times more strongly than formal education — suggesting that the experiential, navigational dimension is exactly what formal curricula miss.

The 52 bills mandate the formal curriculum. None mandate the experiential, communicative dimension. The gap between those two tracks is the ALC Stratification Problem at legislative scale.

What "AI Literacy" Should Actually Mean

A communication-theoretic AI literacy would include:

• Infrastructure awareness — understanding that the interface is a surface over architecture, not the thing itself
• Navigational fluency — the ability to communicate effectively within software's constraints, not just to operate its features
• Repair literacy — knowing what to do when AI breaks down, misunderstands, or produces harmful output
• Agency recognition — identifying when the tool is shaping your request before you've made it (the Agency Inversion problem)
• Stratification awareness — understanding that your experience of a tool is not universal

None of these require abandoning tool proficiency. They require supplementing it with the communicative framework that makes proficiency meaningful rather than merely functional.

The Window Is Closing

Legislative definitions calcify. Once 25 states have codified AI literacy as tool proficiency, changing that definition becomes a multi-year, multi-state advocacy project. The time to inject communication theory into the conversation is now — while the definitions are still being written.

Peter Thiel recently argued that AI is "coming for the math people before the word people." He's wrong, but not in the way you'd expect. The real divide isn't math versus words. It's navigation versus consumption. Those with application layer fluency will compound advantage regardless of their domain. Those without it will be optimized by systems they can't see.

Fifty-two bills across 25 states. A generation of AI literacy policy being written in real time. And the discipline that studies how humans communicate — the one that should be at the center of this conversation — is nowhere in the room.