KPMG Australia Catches Staff Cheating on AI Ethics Exam

KPMG Australia confirmed in February 2026 that 28 staff had used AI tools to cheat on a mandatory internal training exam on the ethical and responsible use of AI, since the firm introduced AI-use monitoring for internal testing in 2024 (Australian Financial Review; The Guardian). The most senior case involved a registered company auditor at partner level, fined more than A$10,000 and required to retake the exam, who uploaded the course's reference manual into an external generative AI tool to produce answers during what was designed as an open-book assessment (Space Daily; NDTV). The partner self-reported the conduct to Chartered Accountants Australia and New Zealand, which opened a separate investigation; the remaining 27 cases involved staff at or below manager level (Space Daily). CEO Andrew Yates told the Australian Financial Review, "As soon as we introduced monitoring for AI in internal testing in 2024, we found instances of people using AI outside our policy. We followed with a significant firm-wide education campaign and have continued to introduce new technologies to block access to AI during testing." KPMG has said it will disclose AI-related cheating cases in its annual results going forward.

• •September 2021: The PCAOB fines KPMG US$450,000 and orders an independent review after a 2016-2020 exam-cheating scandal.
• •October 2024: KPMG Australia becomes the first organization globally to earn ISO 42001 certification for AI management systems.
• •December 2025: The Australian Financial Review first reports on staff using AI to cheat on KPMG's internal AI-ethics exam.
• •February 2026: KPMG Australia confirms 28 staff cheated using AI, including a partner fined more than A$10,000.

KPMG's response followed a familiar escalation: deploy AI tools for staff use, prohibit them during assessments, detect violations after staff used them anyway, then add technology to block AI access during testing outright. Space Daily reports that Australian Securities and Investments Commission (ASIC) oversight of the episode was limited by a reporting rule that only requires a formal filing once a professional body has concluded disciplinary action, which had not yet happened when the story broke; Greens senator Barbara Pocock raised the disclosure gap directly with ASIC at Senate estimates.

Firms running internal AI-literacy or certification programs should treat open-book, take-home assessment formats as effectively unproctored against generative AI unless technical blocking or detection is explicitly built in; policy statements alone did not stop misuse at a firm with a certified AI governance program. Compliance and L&D teams should also plan for how AI-cheating disclosures will be handled in regulatory filings and annual reporting, since KPMG has committed to doing so going forward.

Whether Chartered Accountants Australia and New Zealand's investigation results in a formal finding that triggers a required ASIC filing; how KPMG's annual results disclose the scope of the 28 cases; and whether other professional bodies follow the Association of Chartered Certified Accountants, which is ending remote exams from March 2026 over inadequate AI-cheating safeguards.

This adds to a recognizable Big Four pattern - EY's 2022 US$100 million SEC fine over CPA ethics-exam cheating, KPMG Netherlands' 2024 US$25 million PCAOB fine for exam-sharing, and Deloitte Australia's 2025 refund over a government report containing AI-fabricated citations - in which internal integrity controls have consistently lagged the tools staff actually use. The distinguishing feature this time is that the tool being misused and the subject being tested were the same thing: AI itself.