A major controversy has erupted after a Canadian government healthcare report prepared by Deloitte was found to contain fabricated citations and references to nonexistent academic papers, raising serious questions about the consulting giant’s use of artificial intelligence and quality control processes.
The 526-page report, commissioned by Newfoundland and Labrador’s government for nearly $1.6 million, was meant to guide critical healthcare policy decisions. Instead, an investigation by the Independent has uncovered a troubling pattern of false citations, made-up research papers, and misattributed authorship that appears to be the hallmark of AI-generated content.
Fabricated Research and Phantom Citations Undermine Key Health Report Advising Newfoundland and Labrador
The report was designed to advise the province’s Department of Health and Community Services on pressing issues like virtual care implementation, retention incentives for healthcare workers, and the lingering impacts of the COVID-19 pandemic.
This guidance was particularly crucial as Newfoundland and Labrador grapples with severe staffing shortages among nurses and doctors.
However, the investigation revealed multiple red flags. The report cited academic papers that simply don’t exist, pulled conclusions from fabricated research for cost-effectiveness analyses, and even attributed work to researchers who never collaborated on the studies in question. One particularly egregious example referenced a paper supposedly listing seven coauthors, several of whom had never actually worked together.
“It sounds like if you’re coming up with things like this, they may be pretty heavily using AI to generate work,” said Gail Tomblin Murphy, an adjunct professor at Dalhousie University’s School of Nursing.
She discovered she had been cited in a nonexistent academic paper alongside authors she’d only worked with partially three out of six named colleagues.
Murphy emphasized the serious implications: “We have to be very careful to make sure that the evidence that’s informing reports [is] the best evidence, that it’s validated evidence.
And that, at the end of the day, these reports not just because they cost governments and they cost the public [are] accurate and evidence-informed and helpful to move things forward.”
Deloitte Canada Defends AI Use in Report Amid Citation Errors and Prior Australian Controversy
Deloitte Canada has defended its work while acknowledging some problems. “We are revising the report to make a small number of citation corrections, which do not impact the report findings,” a company spokesperson told Fortune.
The firm claimed that “AI was not used to write the report; it was selectively used to support a small number of research citations.”
Despite the errors, the report remains publicly available on the Canadian government’s website as of this week. The province paid for the study in eight installments, according to documents obtained through an access to information request.
Tony Wakeham, who became the province’s new premier in late October as leader of the Progressive Conservative Party, has not yet publicly addressed the controversy. Neither his office nor the Department of Health and Community Services responded to requests for comment.
This isn’t Deloitte’s first brush with AI-related problems. Just last month, the consulting firm admitted to using AI in a $290,000 report for the Australian government focused on welfare reform. That 237-page study, published in July, also contained fabricated academic citations and even a fake quote attributed to a federal court judgment.
The Growing Crisis of Faulty Reports and Government Oversight
After a researcher flagged these hallucinations, Deloitte quietly uploaded a revised version to the Australian government’s website. The firm acknowledged using Azure OpenAI to help create the report but insisted the corrections didn’t affect the substantive findings or recommendations.
The Australian government required Deloitte’s member firm to provide a partial refund for that flawed report. No information has been released yet about whether Canada will seek similar reimbursement for its $1.6 million healthcare study.
These incidents highlight growing concerns about the use of AI in professional consulting work, particularly when it informs government policy decisions that affect millions of people. While AI tools can accelerate research and analysis, they’re notorious for generating plausible-sounding but entirely fictional information a phenomenon known as “hallucination.”
The stakes are especially high in healthcare policy, where decisions based on faulty evidence could impact patient care, resource allocation, and workforce planning during critical shortages.
As governments and organizations increasingly rely on expensive consulting reports to guide major decisions, these cases raise urgent questions about oversight, quality control, and the responsible use of AI in high-stakes professional work.
