Deloitte will partially refund the federal government after admitting to using artificial intelligence in a $440,000 report that contained several errors.

The Department of Employment and Workplace Relations (DEWR) commissioned Deloitte to review its targeted compliance framework and IT system in December 2024, but the report came back littered with mistakes in the references and footnotes.

The report has since been corrected with deleted references and footnotes, corrections to errors and a new reference list.

A DEWR spokesperson confirmed that the firm will repay the final instalment under its contract over the issue, which will be made public after the transaction is finalised.

University of Sydney welfare academic Chris Rudge was the first to notice more than a dozen mistakes when the report was handed down and published on the department’s website in July. 

The report, which is across his area of expertise, contained made-up references that were attributed to his colleagues.

“I was in no doubt, when I read the names of the works, that they were fake because I work in the area and I know my colleagues’ work,” he said.

“Once I discovered one, I just discovered more and more.”

One of the incorrectly referenced works was titled “The Rule of Law and Administrative Justice in the Australian Social Security System”.

The original work is called “The Rule of Law and the Australian Constitution”.

Rudge said the mistakes contained “all the hallmarks” of AI.

“I cannot understand how a human could create titles of works that don’t exist, that don’t appear on Google. How would they do that?” he said.

“I think it’s good if it’s AI, because to think of a person doing that is almost worse.

“It’s very disrespectful to those who have done the research to just not get it right.” 

The report was corrected in September, after the mistakes were first reported by the Australian Financial Review in August.

An updated report was reuploaded to the website on Friday, noting a “small number of corrections to references and footnotes” that it said “in no way impact or affect the substantive content, findings and recommendations in the report”.

In it, Deloitte also acknowledged that it had used “a generative artificial intelligence (AI) large language model (Azure OpenAI GPT-4o) based tool chain licensed by DEWR and hosted on DEWR’s Azure tenancy” in its methodology.

But the firm, which runs its own AI Institute to advise clients and the industry on AI adoption and fluency, declined to confirm whether AI had caused the mistakes.

Instead, a Deloitte spokesperson simply said: “The matter has been resolved directly with the client.”