Deloitte Cited Researchers Who Didn't Write the Papers in a Government Health Report
What happened
A government health report that names specific researchers and cites specific studies carries the authority of those credentials. Readers, policymakers, and other researchers reasonably assume that the people listed actually wrote what they are credited with. When Newfoundland and Labrador released its 2025 Health Human Resources Plan, the document arrived as commissioned professional analysis with Deloitte's name on it. Then The Independent reported that some of the research it cited apparently does not exist.
The plan, released in May 2025, was designed to guide the province's health workforce strategy, the kind of document that shapes hiring targets, training investments, and service delivery for years. After it became public, The Independent identified citations that appeared to be inaccurate or entirely fabricated. In several cases, the researchers whose names appeared in those citations denied having authored the papers the report attributed to them. People who had nothing to do with the cited work were listed as its authors, lending credibility to claims they never made.
Deloitte's response confirmed what the errors suggested. The firm acknowledged that AI was selectively used to support some of the citation work and said it was revising the report. The acknowledgment is notable for what it sidesteps: no accounting of how many citations AI produced, which conclusions they were meant to support, or what review process was supposed to catch fabricated references before the document reached a provincial government. The phrase "selectively used" does not answer any of those questions, and it does not explain why named researchers had no apparent connection to the work they were credited with.
The stakes here go beyond footnote embarrassment. Health workforce planning is evidence-dependent by design. A recommendation about nursing shortages or specialist recruitment that rests on a citation that was never written by the person named is not a recommendation with a minor formatting error; it is a recommendation with no foundation. The researchers whose names appeared without their consent are also placed in an uncomfortable position, listed as authorities on claims they may not hold and linked to a report they had no part in producing.
The incident reflects a gap that extends beyond this report. When AI contributes to citation work in a professional deliverable, current practice offers no standard mechanism for establishing what the system actually generated, which references were checked by a person, and when. There is no provable record of what a system did, in the way an audit trail would provide, so errors surface only if someone happens to notice. Here, the signal was researchers reading their own names on papers they never wrote. Without that particular form of accidental discovery, the fabrications might have stayed in the published record indefinitely.
Reported impact
- Affected parties
- Not publicly disclosed
- Harm type
- Not publicly disclosed
- Scale
- Not publicly disclosed
- Financial impact
- Not publicly disclosed
- Regulatory action
- Not publicly disclosed
Classification
Relevant governance controls
Governance control mapping is not available for this record.
- No controls mapped
Not publicly disclosed
Control mapping is analytical. It does not state that any control would have prevented the incident.
Sources and evidence
This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.