Tuesday, October 7, 2025

Deloitte to pay money back to Albanese government after using AI in $440,000 report

Partial refund to be issued after several errors were found in a report into a department’s compliance framework

Deloitte to pay money back to Albanese government after using AI in $440,000 report

Deloitte will provide a partial refund to the federal government over a $440,000 report that contained several errors, after admitting it used generative artificial intelligence to help produce it.

The Department of Employment and Workplace Relations (DEWR) confirmed Deloitte would repay the final instalment under its contract, which will be made public after the transaction is finalised. It comes as one Labor senator accused the consultancy firm of having a “human intelligence problem”.

Deloitte was commissioned by the department to review the targeted compliance framework and its IT system, used to automate penalties in the welfare system if mutual obligations weren’t met by jobseekers, in December 2024.

Sign up: AU Breaking News email

The subsequent report found widespread issues, including a lack of “traceability” between the rules of the framework and the legislation behind it, as well as “system defects”. It said an IT system was “driven by punitive assumptions of participant non-compliance”.

It was first published on 4 July. It was re-uploaded to the DEWR website on Friday, after the Australian Financial Review in August reported that multiple errors had been found, including nonexistent references and citations.

Related: Labor asks Deloitte to design universal childcare system as PM eyes political legacy

University of Sydney academic, Dr Christopher Rudge, who first highlighted the errors, said the report contained “hallucinations” where AI models may fill in gaps, misinterpret data, or try to guess answers.

“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,” he 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.”

The updated review noted a “small number of corrections to references and footnotes”, but the department has said there have been no changes to the review’s recommendations.

“Deloitte conducted the independent assurance review and has confirmed some footnotes and references were incorrect,” a spokesperson for the department said.

“The substance of the independent review is retained, and there are no changes to the recommendations.”

In the updated version of the report, Deloitte added reference to the use of generative AI in its appendix. It states that a part of the report “included the use of 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.”

Deloitte did not state that artificial intelligence was the reason behind the errors in its original report. It also stood by the original findings of the review.

“The updates made in no way impact or affect the substantive content, findings and recommendations in the report,” it stated in the amended version.

A spokesperson for Deloitte said “the matter has been resolved directly with the client”.

Rudge said that, despite his criticism, he hesitates to say the whole report should be “regarded as illegitimate”, because the conclusions concur with other widespread evidence.

Labor senator Deborah O’Neill, who was on a senate inquiry into the integrity of consulting firms, said it looked like “AI is being left to do the heavy lifting”.

“Deloitte has a human intelligence problem. This would be laughable if it wasn’t so lamentable. A partial refund looks like a partial apology for substandard work,” she said.

“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.

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

The AFR found several incorrect references in the original report, including nonexistent reports by professors at the University of Sydney and the Lund University in Sweden.

The paper also reported a made up reference to a court decision in a robodebt case, Deanna Amato v Commonwealth. Deloitte wrote in it’s final report that the update “amend[ed] the summary of the Amato proceeding which contained errors”.

The best public interest journalism relies on first-hand accounts from people in the know.

If you have something to share on this subject, you can contact us confidentially using the following methods.

Secure Messaging in the Guardian app

The Guardian app has a tool to send tips about stories. Messages are end to end encrypted and concealed within the routine activity that every Guardian mobile app performs. This prevents an observer from knowing that you are communicating with us at all, let alone what is being said.

If you don't already have the Guardian app, download it (iOS/Android) and go to the menu. Select ‘Secure Messaging’.

SecureDrop, instant messengers, email, telephone and post

If you can safely use the Tor network without being observed or monitored, you can send messages and documents to the Guardian via our SecureDrop platform.

Finally, our guide at theguardian.com/tips lists several ways to contact us securely, and discusses the pros and cons of each. 

Read original article →