Every week, teachers across the country open an AI tool, type a prompt, and receive a lesson plan that looks perfectly aligned to their state standards. The vocabulary sounds right. The learning objectives check the expected boxes. The activities feel appropriate for the grade level.
But nobody has actually verified any of it.
That gap, the space between “sounds aligned” and “is aligned,” sits at the center of a problem that Learning Commons, a nonprofit backed by the Chan Zuckerberg Initiative, is trying to solve. The organization recently launched an open platform designed to give edtech companies and school districts a shared foundation for building AI tools that can actually prove their claims. With more than 70 partners signed on, including OpenAI and Anthropic, the effort is significant enough that every district leader and instructional coach should understand what it means for the tools their teachers use.
The Standards Alignment Problem Is Bigger Than It Looks
Ask any curriculum coordinator what keeps them up at night about AI in classrooms, and “standards alignment” will come up fast. The issue is not that AI tools are generating low-quality content. The issue is that quality and alignment are different things, and AI is very good at producing content that reads well without connecting to the specific, sequenced expectations that a teacher is accountable for meeting.
State standards are not generic. Third-grade math in Texas is not the same as third-grade math in Massachusetts. The language matters. The sequence matters. When a teacher uses an AI-generated lesson without realizing it was built on a generic model of “third-grade math” rather than their actual standards, the lesson can look professional and still miss the mark. By the time a coach or administrator notices the drift, the unit is already over.
Learning Commons is building what it calls open knowledge graphs and AI evaluators, essentially shared infrastructure that edtech companies can plug into so their tools have access to verified, structured representations of state standards. Sandra Liu Huang, president of Learning Commons and a former product leader at Google, Facebook, and Quora, describes this as laying the tracks before the trains start running. The metaphor is apt. Right now, every AI vendor is building their own version of “state standards alignment” from scratch, using different data sources, different update frequencies, and different definitions of what alignment even means. The result is a patchwork that leaves districts without any reliable way to compare tools or verify claims.
Who Decides What Gets Built Into the Foundation?
The open platform framing raises a real question that educators should take seriously: if this infrastructure shapes how AI understands curriculum, standards, and student learning, who controls what goes into it?
This is not a hypothetical concern. The history of edtech is full of products that embedded particular assumptions about what good instruction looks like, then scaled those assumptions across thousands of classrooms before anyone examined them critically. A shared AI foundation, if it becomes widely adopted, could entrench those assumptions at an even deeper level, baked into the tools before teachers ever open them.
Learning Commons frames its answer around openness and governance: the platform is designed so that districts can build their own custom tools on the infrastructure rather than waiting for commercial vendors. That means a district with specific equity priorities, culturally relevant curriculum requirements, or specialized student populations could theoretically shape how the foundation reflects their context rather than accepting a generic national model.
That potential matters. But “theoretically” is doing real work in that sentence. The districts most likely to have the technical capacity to build custom tools are also the districts with larger budgets and more infrastructure. Smaller districts, rural districts, and under-resourced schools will more likely rely on the commercial vendors who choose to build on this foundation. Whether those vendors use the open infrastructure to improve transparency and accuracy, or simply use the Learning Commons affiliation as a marketing claim, will depend heavily on how governance actually works in practice.
Educators should ask hard questions here, not because Learning Commons has bad intentions, but because the stakes are high enough that good intentions alone are not sufficient.
What District Leaders Should Ask Every AI Vendor Right Now
Whether or not Learning Commons’ platform becomes the dominant infrastructure for K-12 AI, Sandra Liu Huang has surfaced a set of questions that every district leader should already be using when evaluating AI tools. They are worth spelling out directly.
First: where does the standards data come from? A vendor who says their tool is “aligned to your state standards” should be able to tell you whether they licensed official standards documents, scraped them from a website, or used a commercial standards database. Those are different things with different accuracy profiles.
Second: how often is the standards data updated? State standards change. Frameworks get revised. A tool built on 2019 standards data will generate lessons that look aligned but miss recent revisions. Update frequency is a basic quality signal that vendors should be able to answer immediately.
Third: is the standard the actual core of the tool, or is it a filter applied at the end? Some AI tools generate content first, then check it against standards as a final step. Others build from the standard outward, treating it as the anchor for every decision. The second approach produces meaningfully better alignment. Asking which approach a vendor uses will tell you a great deal about how seriously they take the problem.
Fourth: can the tool show its work? If a lesson claims to address a specific standard, can the tool surface which part of the lesson addresses which part of the standard? Explainability is an important signal of genuine alignment versus surface-level labeling.
What This Means for Teachers in Practical Terms
For individual teachers, the practical implication is not to wait for infrastructure debates to resolve before using AI tools thoughtfully. A few habits can meaningfully close the verification gap that Learning Commons is trying to address at scale.
When using any AI lesson generator, pull up the actual text of the relevant standard before using the output. Read it yourself. Does the lesson actually address the cognitive demand level the standard requires? A standard asking students to “analyze” requires different work than one asking them to “identify,” and AI tools frequently conflate them.
Build a short verification step into any AI-assisted planning workflow. This does not mean checking every line of a lesson. It means reading the objective the tool generated against the standard text, confirming they are asking students to do the same thing, and flagging quickly when they are not.
When your district evaluates new AI tools, push for pilots that include a standards audit. Ask a curriculum specialist to review a sample of AI-generated lessons before purchasing. The difference between tools that have done this infrastructure work seriously and tools that have not will show up quickly in that audit.
For instructional coaches and curriculum directors, the Learning Commons development is a signal to get ahead of the question your teachers are already asking informally: can I trust this? Building a simple internal rubric for evaluating AI-generated content alignment, before your district is locked into a vendor contract, gives teachers a framework to use and gives leadership a basis for vendor conversations that go deeper than marketing claims.
The Larger Opportunity
Learning Commons’ open platform is ambitious, and the involvement of major AI companies as partners gives it genuine weight. The central problem it addresses, that AI tools are generating curriculum-adjacent content with no reliable way to verify what they produce, is real and consequential. Getting the infrastructure right matters.
But the most important near-term shift has nothing to do with what nonprofits build or what vendors adopt. The shift that helps students now is teachers and leaders becoming better, more specific consumers of the AI tools already in their hands. Understanding what “standards alignment” actually requires, knowing what questions to ask, and building simple verification habits into existing workflows: these are the moves that protect instructional quality while the larger infrastructure debate plays out.
The foundation matters. So does what teachers do while it is being built.
EdTech Institute covers technology in education so teachers can focus on what matters: their students.
EdTech Institute also builds free, classroom-ready tools for teachers at RazaEd.
Source: Who Is Building the AI Foundation for K-12 Classrooms? (EdSurge)