Western Sydney University pro vice‑chancellor Cath Ellis admitted this week that Microsoft’s Copilot was used in the development of an opinion piece published in the Sydney Morning Herald, and the SMH removed the column after inquiries from Guardian Australia.
The university confirmed the admission came only after the story went live; the newspaper ran a mea culpa later the same day and took the item down. Western Sydney University told Guardian Australia that Copilot had been used in producing the piece, but neither the university nor the SMH has published a full breakdown of what Copilot contributed.
The episode landed as Australians are using generative tools in large numbers: Roy Morgan data released this week shows 13.6 million people aged 14 and over — about 58% of that population — use AI each month, with ChatGPT the most popular, followed by Google’s Gemini and Microsoft Copilot. Usage is heaviest among 25‑ to 34‑year‑olds (74%) and 35‑ to 49‑year‑olds (72%). At the same time, a separate survey by the Office of the Australian Information Commissioner last month found just 4% of Australians trust AI and 79% want to be told when it is being used, up from 73% in 2023.
Those numbers explain the friction at the heart of the removal: widespread, growing use on one hand and deep public distrust and a demand for transparency on the other. The immediate question inside newsrooms and universities is not whether AI was used — the university acknowledged that — but how much of the published copy was generated, edited, or shaped by Copilot, and why that was not disclosed before publication.
The gap matters beyond one column. Regulators and institutions are already reacting: Fair Work Australia said this week it will seek new powers to reject applications made with AI, and several academic journals now bar generative AI from contributing to papers. News organizations have lagged behind those formal rules, and the SMH removal spotlights the absence of clear, public standards for disclosing AI assistance in opinion and analysis pieces.
The SMH’s mea culpa did not spell out follow-up steps beyond removing the piece. The most consequential unanswered question is specific and narrow: what proportion of the column was the product of Copilot rather than human drafting or editing? That answer will determine whether this episode becomes a prompt for formal disclosure rules in Australian newsrooms or a one‑off correction.
For readers tracking related coverage and publishing practices, Filmogaz’s earlier note on editorial standards can provide context; see Smh Stock: Why VanEck Semiconductor ETF Appeared in Zacks' Analyst Blog ( Until news organizations set and enforce disclosure rules, instances like this will keep surfacing — and the public, already wary, will keep insisting on knowing when AI has had a hand in what they read.
