Deloitte to pay back after using AI to write $440,000 government report

Deloitte will make a partial payment of $440,000 to the Australian federal government after admitting that generative artificial intelligence (AI) was used to help produce a report riddled with factual and citation errors.

The move follows an internal review that exposed major inaccuracies and prompted public criticism of the consulting giant.

The Department of Employment and Workplace Relations (DEWR) confirmed that Deloitte would refund the final instalment under its contract, with details to be made public once the transaction is completed. The report, commissioned in December 2024, was meant to assess the government’s targeted compliance framework and its IT system, which automates penalties for jobseekers who fail to meet their obligations.

The initial findings, published on July 4, highlighted “a lack of traceability” between the framework’s rules and the underlying legislation, as well as “system defects.” It also said the system was “driven by punitive assumptions of participant non-compliance.”

However, the Australian Financial Review (AFR) later revealed that the report contained several non-existent references and citations. The publication prompted Deloitte to re-upload a corrected version on Friday. University of Sydney academic Dr. Christopher Rudge, who first identified the errors, said the report contained “hallucinations” typical of AI-generated text.

“Instead of just substituting one hallucinated fake reference for a new ‘real’ reference, they’ve substituted the fake hallucinated references and in the new version, there’s like five, six or seven or eight in their place,” Dr. Rudge said. “So what that suggests is that the original claim made in the body of the report wasn’t based on any one particular evidentiary source.”

In the revised version, Deloitte admitted that part of the report was created using a generative AI tool — “Azure OpenAI GPT-4o” — licensed by DEWR. Despite the corrections, both Deloitte and the department maintained that the review’s core findings and recommendations remained valid.

“Deloitte conducted the independent assurance review and has confirmed some footnotes and references were incorrect,” a DEWR spokesperson said. “The substance of the independent review is retained, and there are no changes to the recommendations.”

Deloitte, in its updated statement, also defended the integrity of the report, insisting that “the updates made in no way impact or affect the substantive content, findings and recommendations.” A spokesperson added, “the matter has been resolved directly with the client.”

Labor senator Deborah O’Neill, who sits on a parliamentary inquiry into the integrity of consulting firms, criticised Deloitte’s handling of the project. “Deloitte has a human intelligence problem. This would be laughable if it wasn’t so lamentable,” she said. “A partial refund looks like a partial apology for substandard work. Anyone looking to contract these firms should be asking exactly who is doing the work they are paying for, and having that expertise and no AI use verified.”

She added wryly, “Perhaps instead of a big consulting firm, procurers would be better off signing up for a ChatGPT subscription.”

Dr. Rudge said he does not believe the entire report should be “regarded as illegitimate,” noting that its conclusions align with other independent evidence of systemic flaws in the welfare compliance system.