Adelphi University Sued Over AI Allegation

Orion Newby, a sophomore at Adelphi University in New York, is suing the school, claiming that he was falsely accused of using generative AI on a history paper.

According to the lawsuit, Newby has learning differences and, as part of that, works closely with tutors to improve his grammar. He claims that when he submitted a history paper on Christianity, it was flagged by Turnitin’s AI detection system and deemed “too advanced” by the instructor for a first-year student.

Newby staunchly denies using AI to write the paper and claims to have worked on it for between 15 and 20 hours. Nonetheless, the instructor gave the paper a zero, and the school sided with the instructor.

He claims that things came to a head when he was ordered to take an anti-plagiarism course before enrolling in any new classes. He and his family filed the lawsuit, hoping to get this punishment set aside and the case removed from his record.

Newby isn’t the first student to file a lawsuit alleging that they were falsely accused of using AI. In January, Haishan Yang, a Ph.D. student at the University of Minnesota, filed a lawsuit against his school. He is seeking over $1.3 million in damages.

Schools should be acutely aware of the escalating number of these lawsuits. The potential legal risks for schools are significant. However, there are proactive steps they can and should take to protect themselves.

The Good News and the Bad News

For universities, this is a good news/bad news situation.

The good news is that, in the United States, courts are generally reluctant to intervene in areas of academic integrity. As long as students have due process and there is no breach of contract or allegations of discrimination, courts largely leave matters of academic integrity to the experts, the schools themselves.

The bad news is that AI presents new challenges in this space. While a school or expert can review a traditional plagiarism analysis, AI detectors are a black box. It makes a decision, and no one (not even the tool’s creators) can definitively say why.

Though some AI detectors have tried to mitigate this by adding explanations about why specific passages were flagged, there’s no way to see the process under the hood and check the detector’s work.

The black box nature of AI detectors means that even a small number of false positives can significantly undermine the tool’s credibility. This underscores the crucial need for additional verification to avoid false positives.

However, obtaining that verification can be challenging. The ‘Swiss Cheese’ approach, in which educators use multiple layers to prevent and detect AI use, can provide some protection.

Still, the nature of AI detectors makes them vulnerable to legal challenges. Students can argue that they didn’t receive due process because they couldn’t cross-examine or rebut the most critical piece of evidence.

While it is unclear how courts will respond to these claims, AI opens new arguments students can use to try to coax courts into intervening. The success of those attacks will likely depend on how schools use the AI detector and what other evidence they have.

Schools need to be careful not to put themselves in that position.  

Protecting Schools and Educators

This threat raises a simple question: What should schools and educators do?

First, all the advice I collated in January still holds. Creating a less combative environment around plagiarism, including AI, using the “Swiss Cheese” approach to mitigate AI usage, and encouraging students to use tools that track their writing are all positive steps.

But as more and more cases head to court, it’s also essential to consider the legal issues. How are students accused of AI wrongdoing going to receive due process? What other ways can you verify (or disprove) the usage of AI?

These are not easy questions. However, it’s important to remember that AI detectors do not accuse students of AI plagiarism; humans do that. The issue arises when humans rely solely on the AI detector and surrender their own judgment to it.

Getting additional evidence will be crucial. This can be as simple as talking with the student about the assignment and seeing how knowledgeable they are about the topic.

However, document versioning will likely be one of the most powerful tools, as it shows how a document was written over time. It provides physical evidence about whether AI was used. This can be taken to another level with controlled writing environments, such as Grammarly Authorship and Turnitin Clarity.

Simply put, having this additional evidence isn’t just about ensuring that innocent students aren’t punished; it’s about limiting legal liability and heading off legal challenges.

Bottom Line

Looping back to Newby’s lawsuit, this should be an easy case to prove one way or another. He claims to have spent 15-20 hours on this paper and worked with tutors in the university’s Bridge Program.

These should be easy things to prove or disprove. You can talk to the tutors, view drafts of the paper and see how it was written from various witnesses.

While this wouldn’t eliminate the possibility of AI polishing, it would show that the bulk of the work and information came from Newby himself. That would put any AI usage in context, if not eliminate it as a possibility.

But what should worry universities is that this lawsuit is over a relatively minor punishment. A zero on an assignment and requiring the student to take an anti-plagiarism course are relatively tame repercussions.

Newby wasn’t in danger of being expelled, suspended, losing his degree or anything other than a possible hit to his GPA. To make matters worse, the allegations also occurred during his first year as a student. The stakes could barely be any lower.

If this can generate a lawsuit, then any AI-related case can. Schools and educators need to recognize this and start creating AI policies with it in mind.

It’s for the school’s protection as much as it is for the students.