A partner at KPMG Australia has been fined A$10,000 (£5,200) for using an artificial intelligence tool to cheat on an internal training course about AI, prompting questions about ethics and oversight in the age of machine learning. The individual uploaded training materials from the course into an AI platform to generate answers for exam questions, a move detected by KPMG’s monitoring systems in August last year. Following an internal investigation, the partner was required to retake the module.
This incident is part of a growing trend, with KPMG Australia identifying 28 cases of employees using AI to circumvent training protocols since July. The firm, which oversees over 20,000 internal exams annually, uses open-book knowledge checks to assess employee understanding. While staff can reference downloaded materials during exams, uploading them to AI tools is strictly prohibited.
Andrew Yates, CEO of KPMG Australia, told The Australian Financial Review that monitoring AI misuse has been “a very hard thing to get on top of given how society has embraced it.” He added that since introducing AI monitoring in 2024, the firm has identified multiple breaches and launched a comprehensive education campaign, alongside technology to block AI access during testing.
The case has drawn criticism from politicians, including Australian Greens senator Barbara Pocock, who labeled the incident as “another example of unethical behaviour” and called the reporting system “inadequate.” The Australian Securities and Investments Commission confirmed it is in contact with KPMG, which is voluntarily sharing information about AI-related misconduct.
KPMG Australia has pledged to report AI cheating cases in its annual results and to ensure employees are self-reporting violations, as the accounting industry faces increasing scrutiny over ethics and transparency in the era of artificial intelligence.
