The AI Trust Problem in Classrooms Goes Both Ways Now

For the last two years, the story was simple: students use AI to write papers, teachers try to catch them. That story just flipped.

Students are now filing complaints about their professors. At Northeastern, one student found out her professor had secretly used AI to generate class notes, then filed a formal complaint and demanded her tuition back. At Wharton's Executive MBA program, students raised the same issue in a town hall: repetitive, generic feedback on assignments that looked exactly like what an AI grader produces, while the program's AI-disclosure rules applied to students but not to faculty.

Marc Watkins, who studies AI in higher education at the University of Mississippi, put the actual nightmare scenario plainly: students using AI to write the paper, professors using AI to grade it. Neither side is reading the other's work anymore.

The irony nobody planned for

GPTZero built its entire business detecting AI-written student essays. It just launched a product that helps teachers grade with AI. The company that made its name catching AI usage is now selling the other side of it.

Teachers have their own trust problem

Adoption jumped fast: 60% of US public school teachers used an AI tool during the 2024-2025 school year, nearly double the year before. But 70% say they don't feel prepared to use it well, and over 90% want more support than they're getting. More than half of teachers report AI made them more suspicious of their own students' work, turning grading into detective work instead of teaching.

That suspicion has real consequences when detection tools get it wrong. A University of North Georgia student lost her scholarship after Grammarly, the same tool her university recommended, got flagged as AI-generated content. She wasn't cheating. She was using a spell-checker her school told her to use.

The same shift is showing up on the teaching side too

A high school English teacher writing in the Philadelphia Inquirer this month described what's changed for her personally: reading student essays used to bring genuine discovery, now the first instinct is running detection software before reading for meaning at all. It's the same suspicion reflex playing out on both sides of the desk, at the same time students are losing scholarships to false positives from that exact instinct.

What this means if you're picking classroom AI tools

The pattern isn't that AI is bad for education. Trust breaks down in both directions when nobody's transparent about how they're using the tool. A grading tool that gives generic, copy-paste feedback erodes student trust the same way an over-aggressive detector erodes teacher-student trust. Both treat AI output as ground truth instead of a first draft that still needs a real person's judgment.

The question worth asking isn't whether a tool saves time. Ask whether it makes it easier for a human to make the final call, or whether it quietly replaces that call. Tools built around the first idea tend to hold up. The ones built around the second are the ones showing up in tuition-refund complaints.

Sources:

• Northeastern University AI-notes complaint and tuition refund demand: Fortune, via Yahoo
• Wharton Executive MBA AI grading concerns: The Daily Pennsylvanian
• GPTZero's pivot to AI grading tools, Marc Watkins commentary
• Teacher adoption and preparedness statistics: Walton Family Foundation, 2026; Congressional testimony, Rep. Kevin Kiley, Feb 2026
• University of North Georgia scholarship/false-positive detection case: The Independent, reported via AOL
• Teacher perspective on AI and classroom trust: The Philadelphia Inquirer, Aug 19, 2026