Finkel demands strict AI disclosure standards for Australian media and universities

Former university chancellor Alan Finkel has called for strict disclosure standards regarding the use of artificial intelligence in Australian media and higher education, following a controversy involving an opinion piece published by the University of Western Sydney. In an article published in The Guardian, Finkel argued that readers must be explicitly informed if content is AI-generated, citing a recent incident where Cath Ellis, the university’s pro vice-chancellor for quality and integrity, published a rebuttal without disclosing AI involvement.

The controversy began after Dr Kylie Moore-Gilbert of Macquarie University published an opinion piece in the Sydney Morning Herald expressing concern over excessive student use of AI chatbots. In response, Ellis published a rebuttal in the same publication. However, readers identified telltales of AI phraseology in Ellis’s article, leading to significant social media backlash. The University of Western Sydney subsequently acknowledged that Microsoft Copilot had produced early drafts and provided editing, structure, and language refinement, describing the use as sophisticated and appropriate.

Finkel, writing through his organisation Proudly Human, described the university’s defence as flawed. He argued that opinion pieces are exercises in persuasion where readers seek to weigh the insights and prose of the specific author. He stated that while AI may be used for research, spellchecking, grammar checking, and formatting, it must not write sentences or paragraphs. These boundaries, he said, constitute the “de minimis standards” he is advocating for.

In response to the incident, the Sydney Morning Herald and The Age removed Ellis’s article and stated they would require future contributors to guarantee that AI has not been used to write or construct articles. Finkel welcomed this policy but urged mastheads to publish their specific standards and implement technological verification of human authorship where necessary. He warned that without such conditions, human authorship risks becoming irrelevant.

Finkel also criticised the slow institutional response from universities, pointing to the Castlereagh statement issued earlier this year by representatives from most Australian universities and educational associations. He noted that while the statement articulated goals for an AI-transformed future, it lacked specific rules or rapid timelines. He concluded his opinion piece by stating that AI had no role in its drafting or writing.