Your shopping cart is empty!
A Massachusetts school district published AI rules before Big Tech did
Publication date: 18.09.2026
In brief
- Greg Jarboe urges brands to publish an AI accountability document, pointing to a Massachusetts school district that did so before Big Tech.
- Such a document should be dated, name a reviewer and say clearly whether customer data feeds model training.
- The author argues that a published primary document is exactly what AI cites when answering questions about a company's responsibility.
Big Tech spent years on a "build first, ask permission later" playbook, and Greg Jarboe argues in Search Engine Journal that it has stopped working for AI. He cites polling: 61% of Americans oppose new data centers in their neighborhood and 68% think AI regulation is too weak. The gap between use and trust is plain: people open assistants daily while distrusting the companies behind them.
79%of students in the school district understand when generative AI use would undermine learning, and another 72% are clear on what is permitted.
What the school district in Arlington did
The Arlington, Massachusetts school district (ABRSD) published its "AI Guidelines & Guardrails" months before major tech companies produced anything substantive. The document rests on five principles: Humans First, Adaptive Literacy, Responsible Stewardship, Rigorous Governance and Intentional Use. The student survey numbers are telling too: 79% understand when AI would undermine learning and 72% know what exactly is allowed. So the text gets read, and it works.
What interests me most is that an organization with no corporate budget managed what platforms with billion-dollar revenues did not. Trust here is built with a document you can point to.
Three elements a brand can lift onto its site
Jarboe picks out three components, each with its own logic for search.
- An indexable policy document: dated, signed, with names, instead of a vague "we care about ethics". Such a text becomes a primary source that AI can cite.
- Disclosure of AI involvement with a named reviewer. The author calls it a potential trust signal for readers and for the models now citing your content, and warns that it is not a legal hedge for the footer.
- A direct answer on data: whether customer data feeds third-party model training. After Annenberg's research, in his words, people are done taking that on faith.
The practical part the original lacks concerns smaller markets. Companies in Ukraine are only beginning to write such pages, which means there is almost no competition for queries like "does the company use your data to train AI". A page on its own URL with an update date and an author is easy to find and cite. We wrote about who on a team should own such decisions in a piece on the owner of AI marketing accountability, and about making a page machine-friendly in one on a single source of truth for AI.
Where the argument is weaker
Keep in mind that this is an opinion column rather than a study. The author does not show that AI actually cites accountability documents more often than other pages; that is an assumption that sounds logical but nobody has measured. A school district is also no brand: it has no competitors, no sales and no pressure to water down the wording. And it is Jarboe who sums it up sharply, saying Big Tech "keeps trying to win it with a pitch instead of a document".
I am curious how many companies will follow the advice only on paper: publish the page and never name the reviewer.
