The Big Four firm is set to repay the final instalment of its A$439,000 (c. £215,000) contract for an “independent assurance review” of a welfare system. Australia’s Department of Employment and Workplace Relations confirmed on Monday that the report, commissioned in December last year, contained incorrect footnotes and references.

An investigation by the Australian Financial Review in August revealed the document cited non-existent academic reports. While a corrected version of the review was uploaded last week, the government noted that the substance of the findings and its recommendations remained unchanged, Financial Times reports.

This episode highlights the growing risks for consultancies using generative AI, particularly the danger of “hallucinations”, where the technology produces false information. The Big Four firms have invested billions in AI to accelerate their services and maintain an edge over smaller competitors. However, the Financial Reporting Council warned in June that these firms were failing to track how AI affects the quality of their work.

In the updated version of the report, Deloitte added a disclosure to the appendix, stating that part of the review “included the use of a generative artificial intelligence (AI) large language model (Azure OpenAI GPT – 4o)”. While not explicitly blaming AI for the mistakes, Deloitte acknowledged the corrections to citations and references.

In the amended document, Deloitte stated: “The updates made in no way impact or affect the substantive content, findings and recommendations in the report”.

A spokesperson for Deloitte Australia added: “The matter has been resolved directly with the client.”