Deloitte is facing renewed questions over the use of artificial intelligence and quality controls in publicly funded research after a healthcare workforce report it prepared for a provincial government in Canada was found to contain allegedly inaccurate or non-existent citations.
The 526-page Health Human Resources Plan, commissioned by the government of Newfoundland and Labrador for nearly $1.6 million, has come under scrutiny after Canadian news outlet The Independent reported that several references cited in the document could not be traced to real academic papers. The report was intended to guide long-term policy on recruitment, retention and workforce planning in a province battling persistent shortages of nurses and physicians.
According to the publication’s findings, at least four citations referenced in the Deloitte-authored study either do not exist or were inaccurately attributed to real researchers. In some instances, papers were allegedly credited to academics who said they had never worked on them, while other references pointed to research that could not be located in journal databases.
One of the disputed citations attributed to University of Northern British Columbia professor emerita Martha MacLeod was used to support the claim that monetary recruitment incentives create cost savings compared to hiring and training new staff. MacLeod reportedly described the reference as “false” and “potentially AI-generated”. Another cited paper, described as a study on the cost-effectiveness of local recruitment and retention strategies, was said by listed co-author Gail Tomblin Murphy of Dalhousie University to “not exist”.
The Independent also identified references supposedly drawn from the Canadian Journal of Respiratory Therapy that could not be found in the journal’s records.
Deloitte defends findings, confirms revisions
Deloitte Canada has rejected suggestions that artificial intelligence was used to write the report, while acknowledging that amendments are being made.
“Deloitte Canada firmly stands behind the recommendations put forward in our report,” a spokesperson told Fortune, adding that the firm is revising the document “to make a small number of citation corrections, which do not impact the report findings.” The firm said AI was “not used to write the report” but was “selectively used to support a small number of research citations.”
The study, commissioned by the previous provincial government and paid for in eight instalments, examined issues such as virtual care, the impact of the COVID-19 pandemic on healthcare workers and strategies for addressing workforce shortages.
Second AI-linked incident this year
The Canadian controversy follows a similar episode earlier this year involving Deloitte Australia, where a $290,000 government-commissioned report on welfare compliance was found to contain fabricated academic references and a fictitious judicial quote. That report was later revised, and Deloitte Australia provided a partial refund to the federal government. The revised version also acknowledged the use of Azure OpenAI in preparing the document.
While Deloitte maintains that its use of AI is limited and carefully controlled, the episodes in Australia and Canada have intensified scrutiny over how consulting firms deploy emerging technologies and whether existing quality assurance mechanisms are equipped to catch errors before publication. At present, there has been no confirmation on whether any refund will be considered in the case of the Newfoundland and Labrador report, as per media outlets.
