Universities pull back on AI detection tools over accuracy worries

In short: More universities are restricting or disabling AI detection tools because they can wrongly flag student work as AI-written.

What's going on

Universities have been using AI detection tools to try to spot when students use chatbots like ChatGPT to write assignments. These tools give a score that suggests how likely it is that text was written by AI, based on patterns in the writing (a bit like a metal detector that beeps when it thinks it found something).

Concerns are growing that these tools are not reliable enough to use in disciplinary cases. The Financial Times highlights one example involving Orion Newby, a student at Adelphi University, who was accused of using AI after Turnitin produced an “AI-generated score of 100 per cent.” A New York court later ruled in his favour, saying the university failed to follow its own procedures and did not give him a meaningful appeal.

Several universities have since restricted or disabled AI detection features, including Vanderbilt, Yale, Johns Hopkins, Northwestern University, the University of Waterloo in Canada, the University of Cape Town in South Africa, and Curtin University in Australia. Experts cited in the report warn that the biggest risk is false positives, meaning the tool says “AI wrote this” when it did not.

The shift is happening as student AI use rises. A survey of more than 6,600 students across seven UK universities found 32 per cent admitted some unpermitted AI use in assessments.

What to watch

More universities may move away from “surveillance-style” checks and redesign assessments instead. That can include more oral exams, drafts that show the writing process, and clearer rules on when AI help is allowed. Policies are still inconsistent, and research found over 40 per cent of UK universities had no publicly accessible AI policy, which can leave students unsure about what is safe to do.

Source: Financial Times