A University Official Used AI to Write About AI—Without Telling Readers

Cath Ellis runs the integrity office at Western Sydney University. Her job is to make sure academic work is honest and students don't cheat. Recently, she wrote an opinion piece for The Sydney Morning Herald arguing that universities should use artificial intelligence.

Here's the unusual part: she used AI to write that piece, but didn't tell the newspaper or readers that she did.

According to The Sydney Morning Herald, Ellis fed 40,000 words of her own research into Microsoft's Copilot—an AI system—and used it to create the opinion article. Neither she nor the university told the newspaper what happened.

What the University Says

Western Sydney University defended Ellis's choice. University officials said her approach showed "edge thinking and innovative approach" and was a way to prepare students for a world where AI is common.

The university confirmed that AI wrote the article and that it drew from Ellis's earlier research work.

But this raises a question: why doesn't the newspaper deserve to know this? In academic publishing, researchers are required to disclose conflicts of interest, where their funding comes from, and how they did their work. Yet these rules seem fuzzy when academics write for newspapers instead of journals. The newspaper and its readers were left in the dark about how the piece was made.

Who Is Cath Ellis?

Ellis became Pro Vice-Chancellor for Quality and Integrity at Western Sydney University in November 2024. Before that, she worked at several universities, including the University of Wollongong, the University of Sydney, and UNSW.

In 2019, the magazine Times Higher Education named Ellis one of their People of the Year. They recognized her for her work fighting academic dishonesty and preventing cheating. This background makes the situation notable. A specialist in academic honesty used AI without telling anyone—the exact kind of transparency problem she has spent her career preventing.

Why This Matters Beyond One Article

Ellis's situation highlights a tension that technology has created faster than our rules could adapt. What should count as "writing something yourself"?

In this case, Ellis didn't ask AI to generate something from scratch. She gave it 40,000 words of her own research and asked it to reorganize and combine that material. This is different from simply asking AI to answer a question. It's more like handing a personal assistant decades of your notes and asking them to write a summary based on what you've already created.

The problem is that journalism and academic publishing have different expectations about transparency. Academic journals increasingly require authors to say whether they used AI. Newspapers often have looser standards. Neither Ellis nor the newspaper had a clear rule about this, so the gap went unfilled.

The deeper puzzle here is about what "authorship" means when machines help. If you've written the source material and the AI just combines it, is that really AI-generated content? Or is it still yours? These questions don't have easy answers yet.

The Bigger Picture

Universities are adopting AI tools very fast—faster than they're writing clear rules about how to use them. Ellis's case is an early test of whether institutions will be honest about AI assistance, or whether they'll quietly normalize it.

The timing adds irony. An academic whose entire career has focused on maintaining standards and preventing unauthorized help just became a case study in disclosure ethics. Her own field is now watching to see how this gets handled.

For other universities and publishers, this case highlights a gap. Do we need new, clearer rules about when and how to disclose AI use in opinion pieces, news articles, and public commentary? Right now, no one has a standard answer.

What Comes Next

This incident is likely to push universities and media outlets to write clearer policies. They'll need to decide: When someone uses AI, who needs to know? Readers? Editors? Both? And when an academic with Ellis's background is involved, what's the standard?

One thing that might become clear is that different types of AI use might need different rules. Asking AI to write something from scratch is different from asking it to synthesize your own previous work. As these tools become more common in academic life, we may need more specific disclosure standards that capture these differences rather than simple yes-or-no questions about whether AI was involved.

For now, Ellis's case serves as a reminder that the gap between how fast technology moves and how fast institutions can adapt to it remains wide. The university world has to decide what transparency looks like in an age when machines can write and synthesize. How they answer will shape what happens next.