Kelley School of Business Banned AI Detectors — Calls Them ‘Highly Unreliable’

This isn’t a rogue professor’s op-ed. It’s a formal policy document from one of America’s top business schools, and it names names. The playbook lists the tools by name and declares they produce both false positives and false negatives at rates that make them unfit for academic enforcement. The reasons go further than accuracy: data security, intellectual property risk, and — critically — documented bias against non-native English speakers.

That last point matters enormously. Studies on AI detector false positives have shown that ESL writers are flagged at dramatically higher rates than native speakers, even when writing entirely without AI assistance. A Stanford-linked study found some detectors flag non-native English essays as AI-written up to 61% of the time. Kelley didn’t ban detectors because of a vague concern about accuracy — they banned them because those tools fail in ways that are discriminatory and legally risky.

Kelley isn’t alone. According to our own resource tracking universities that have banned AI detectors, over 50 institutions have now moved away from automated AI detection — but most did so quietly, disabling Turnitin’s AI feature without much fanfare. What makes Indiana’s move different is that they published a playbook. They didn’t just stop using the tools; they told other institutions exactly why and what to do instead.

The “instead” is the other shoe. Kelley is pushing professors to redesign assignments around reasoning, judgment, and process — work that requires students to demonstrate understanding rather than submit a polished output. It’s sound pedagogy. It also signals something the detection industry doesn’t want to admit: the arms race was always unwinnable, and smart institutions are opting out rather than escalating.

The obvious counterargument is that banning detectors gives students carte blanche to use AI without accountability. That’s a reasonable concern — but it misunderstands what detectors actually provided: the appearance of accountability, not the substance of it. A tool that flags innocent students and misses guilty ones at comparable rates isn’t a deterrent; it’s a liability. Kelley’s answer is assessment redesign, not amnesty — and that’s the right call.

For students who’ve been humanizing AI-written text to protect themselves from false positives, this is a strange kind of validation. The tool you were defending yourself against has just been declared unfit for purpose by a major university. The arms race isn’t over — plenty of schools still use these tools — but the institutions running the math are quietly walking away. Indiana just did it loudly.