In California, Colleges Pay a Steep Price for Faulty AI Detectors [ Undark ]

The flava:

It has been more than two years since the release of ChatGPT created widespread dismay over generative AI’s threat to academic integrity. Why would students write anything themselves, instructors wondered, if a chatbot could do it for them? Indeed, many students have taken the bait, if not to write entire essays, then certainly to draft an outline, refine their ideas or clean up their writing before submitting it.

And as faculty members grapple with what this means for grading, tech companies have proved yet again that there’s money to be made from panic. Turnitin, a longtime leader in the plagiarism-detection market, released a new tool within six months of ChatGPT’s debut to identify AI-generated writing in students’ assignments. In 2025 alone, records show the California State University system collectively paid an extra $163,000 for it, pushing total spending this year to over $1.1 million. Most of these campuses have licensed Turnitin’s plagiarism detector since 2014.

That detector first became popular among professors when the internet made it easy for students to copy and paste information from websites into their assignments. In the AI detector, faculty members sought both a way to discourage students from using ChatGPT on their homework and a way to identify the AI-generated writing when they saw it.

But the technology offers only a shadow of accurate detection. . . .  

The article:

https://undark.org/2025/07/01/california-colleges-ai-detectors/