A healthcare workforce report commissioned by the Canadian province of Newfoundland and Labrador has prompted a formal review after it was found to contain several incorrect academic citations.
The document, delivered by Deloitte, was created to guide the province’s healthcare staffing strategy over the next decade and was released in May 2025. However, multiple experts identified references to non-existent studies and misattributed authors, prompting an investigation.
This is not the first time the consulting giant has been accused of including fabricated research in government-commissioned work.
Deloitte acknowledged the citation issues, confirming that four specific references were incorrect. The firm stated that the errors were limited in scope and did not affect the overall conclusions of the report. AI tools were used selectively to support parts of the research, according to a Deloitte spokesperson, though the extent and context of this usage remain under review.
Why It Matters: Organizations that rely on external consulting partners for research and strategy documents increasingly face new challenges as AI tools become part of the production process. As businesses and governments partner with firms for transformation, staffing models, and AI-driven insights, the need for rigorous verification and audit trails becomes essential.
- Fabricated Sources Embedded in a Critical Government Report: The 526-page healthcare report contained false academic citations, including references to nonexistent studies and incorrect attributions to real researchers. The report was intended to inform long-term staffing decisions during a period of healthcare labor shortages. Its compromised credibility shows how easily unverified information can be embedded into high-value deliverables and used to justify major investments or policy moves.
- Generative AI Use Not Originally Disclosed: Deloitte initially denied the use of generative AI in producing the report, later admitting that it was used to support parts of the research. The lack of clear disclosure raises concerns about transparency in vendor methodologies. When AI tools are used to assist in generating or sourcing content, without appropriate human oversight or verification, the risk of introducing unreliable data into documentation increases significantly.
- Reputational Risk Emerges Alongside Financial Waste: The healthcare report is now being reviewed, and public calls for reimbursement have been made, echoing a similar case in Australia where Deloitte issued a partial refund after another AI-assisted report included hallucinated content. Inaccurate deliverables not only waste funds but can also damage institutional trust and provoke public or internal backlash, especially in sensitive industries.
- Vendors Face Growing Pressure for Evidence-Based, Auditable Work: With two high-profile incidents involving the same firm, expectations around accountability are shifting. Organizations are increasingly expected to hold vendors to strict evidence standards requiring documentation, verifiability, and clearly defined research methods. Deliverables that cannot withstand evaluation may lead to project delays or legal and financial consequences.
- Stronger Contractual and Review Safeguards Are Becoming Essential: To avoid reliance on flawed inputs, more contracts now need to include specific language addressing the use of AI in research and post-delivery review rights. Where vendors are relied upon for insights or operational planning, the need for enforceable quality assurance mechanisms has never been more urgent.
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
