OpenAI Just Put ChatGPT in Front of 300,000 Educators. The Number That Matters Is 900,000.

On August 26, 2026, OpenAI announced that ChatGPT for Teachers is expanding to 55 additional school systems across 20 states.

The product now reaches more than 300,000 educators and staff across 100+ K–12 organizations in 30 states. It stays free for verified U.S. K–12 educators through June 2028. A 16-state data privacy agreement gives districts a path to adopt it without negotiating one contract at a time.

That is a big footprint. But the most revealing part of the announcement is a usage number.

Between January 1 and July 16, 2026, educators sent 1.9 million time-saving-task messages through the product. Of those:

• 900,000 were about report cards and progress reports.
• 800,000 were about lesson planning.
• 100,000+ each were about substitute plans and evaluation materials.

Those 1.9 million messages are OpenAI's count of time-saving tasks, not every teacher message in ChatGPT. Even so, report cards were the largest slice of that work. Teachers are asking for help telling someone how a student is doing.

What teachers are really asking for

Report cards are the moment when a teacher must turn months of scattered evidence into a clear statement about a child.

That is hard, slow work. It is also the work most likely to be rushed, because it happens for 25 or 30 students at once, under a deadline.

So teachers are doing something sensible. They are handing a language model their notes and asking it to write the summary.

The output looks fine. The sentences are clean.

But the input is the problem.

A language model can only describe the evidence the teacher already has. If the teacher's evidence is a gradebook of percentages, the report card will be a gradebook of percentages in prose form. "Struggles with fractions" becomes "continues to develop foundational understanding of fractions." Same information, more words.

The gap nobody is filling

Here is the sequence a report card should actually rest on:

Assessment evidence → specific misconception → intervention attempted → did it work → what comes next.

Most schools have the first step and the last step. The middle is missing.

A score of 62% does not say whether the problem was equivalent fractions, unlike denominators, integer rules inside a multistep expression, or a prerequisite gap from two grades earlier. A chatbot asked to "write a progress comment" cannot invent that diagnosis. It can only make the absence of it sound polished.

This is the difference between AI that writes about learning and AI that understands what happened in the classroom.

Gallup found that 58% of teachers have no guidance on using AI for grading and feedback. Weekly users still save hours on production work. Neither fact means the comment is more accurate.

Why grading and reporting is becoming a policy issue

Schools are starting to notice.

Oklahoma's S.B. 1734, enacted in 2026, prohibits AI from being the sole basis for grading or discipline decisions.

So one of the largest teacher-facing uses is also one of the least directed, and one of the most legally watched. That is not a stable situation.

A different design for the same problem

The teacher's need is legitimate: turn evidence into a clear, defensible statement about each student, fast.

The question is what the evidence looks like when it reaches the teacher.

At Omnix360 we designed the loop to run before the report card, not at it:

Plan → Teach → Diagnose → Verify.

By the time a progress report is due, the teacher already has topic-level evidence per student: what was mastered, what was retaught, and what still needs work. The parent dashboard shows the same picture, continuously, instead of once per term. The summary is easier to write because the diagnosis already exists.

AI recommends, the teacher decides. The student never interacts with the AI. And the report card stops being the first time anyone looks closely.

See what parents actually need in a student progress report and teacher-facing AI and misconception radar.

The takeaway

OpenAI's expansion confirms something the industry has been slow to admit: teachers want AI for the work behind the classroom, not in front of the student.

900,000 report-card messages are a demand signal. They say teachers are drowning in evidence they cannot easily interpret.

Writing the comment faster is a relief. Knowing what the comment should say is the real product.

See Teacher-Facing AI in Action.

Sources

• OpenAI: Bringing ChatGPT for Teachers to more U.S. school districts (August 26, 2026)
• Gallup: Most Teachers Receive No Formal Guidance on AI Use (May 26, 2026)
• FutureEd: 2026 State AI in Education Bills tracker (Oklahoma S.B. 1734)