A disputed method, taught nationwide; Claude for Teachers
Anthropic just gave every verified K-12 teacher in America free access to a premium version of Claude that’s mapped to “academic standards in all 50 states.” When Claude for Teachers helps with a lesson, it does not invent academic standards. It reaches into a specific, public dataset and pulls from it. The dataset is called the Learning Commons Knowledge Graph, a Chan Zuckerberg Initiative project that Claude for Teachers connects to rather than one Anthropic built. It is open under a Creative Commons license, and I analyzed the entire thing, all 222,865 standard items spanning the 50 states, Washington, D.C., and the national frameworks they share.
I read it against one question, the only one that matters to a parent. Is this teaching a fact or a skill, or is it teaching a contested claim as though it were settled? What I found was three completely different kinds of bad in here, and the difference between them is not severity. It is reach. To see it, you first have to understand where these standards come from.
Where the 222,865 come from
Start with the shape of the pile, because nobody explains it and it changes everything.
At the bottom are a small number of national frameworks that almost every state builds on. Common Core supplies the math and the English. The Next Generation Science Standards, NGSS, supplies the science. A framework called C3, short for College, Career, and Civic Life, supplies the skeleton for social studies. These are the source documents, and they are compact: the Common Core math list is about 800 statements, its English list about 2,400, NGSS about 1,300, and C3 about 450.
On top of that national base sit 208 separate state frameworks, one per subject per state. And here is the part that matters. In math, science, and English, the states mostly copy the national base. When I count how many states share the exact same standard text, the most-adopted lines in the whole dataset are pure skills lifted straight from the national frameworks: “Reason abstractly and quantitatively” appears in 39 states, “Science and Engineering Practices” in 40, a trigonometry standard about sine and cosine in 41. That shared national skeleton is where the country actually agrees.
Social studies is the exception, and it is the crucial one. C3 is only a skeleton of skills, not content. It never says what history to teach. So in social studies every state writes its own standards from scratch, which is why 94% of all unique standard texts in the entire dataset turn out to be one-state originals that no other state adopted. Social studies is where a state gets to say what it wants, in its own words, accountable to no shared national text.
When you find something ugly in this data, the very next question is how many states adopted it. If the answer is one, you are looking at a local capture. If the answer is thirty, you are looking at something the national frameworks handed to everyone.
So let us sort the bad by reach.
