Two recent government-commissioned reports — one for the Canadian government and another for the Australian government — have been found to contain fake academic citations, nonexistent sources, and in one case, a fabricated judicial quote. These alarming findings raise questions about the reliability of AI-assisted consulting work, and demand urgent scrutiny of how generative AI is used in high-stakes government projects.
What Happened: The Scandals in Canada and Australia
🇨🇦 Canada Healthcare Report — Fake Citations and Phantom Papers
- A 526-page healthcare report by Deloitte for the province of Newfoundland and Labrador in Canada, valued at about CAD 1.6 million, has been flagged for containing multiple fabricated academic citations.
- The report included fake research papers — sometimes listing real researchers as co-authors of studies they never conducted. Some citations referenced articles in legitimate journals that, upon inspection, do not exist.
- According to one academic cited in the fake references, she had never collaborated with the other purported co-authors.
- Despite the bogus citations, Deloitte Canada maintains that the report’s overall recommendations and conclusions remain valid. The firm claims that only a “small number” of references will be corrected.
🇦🇺 Australian Government Report — AI “Hallucinations” Lead to Refund
- Earlier in 2025, Deloitte’s Australian arm prepared a 237-page report for the Department of Employment and Workplace Relations (DEWR). The report cost AU$440,000 (approx. US$290,000).
- After a researcher flagged numerous problems — including references to non existent academic papers and a fabricated quote attributed to a federal court judgment — the firm admitted the mistakes.
- Deloitte Australia agreed to partially pay back the government fee and republished a revised report with the flawed references removed. The revised version disclosed that it used a generative AI tool (Azure OpenAI GPT-4o) during drafting.
- Critics call the incident a clear example of “AI hallucinations” — where generative AI confidently invents plausible-looking but false content.
Why This Matters — Risks of AI-Assisted Consulting
⚠️ Erosion of Trust in Professional Reporting
These incidents have severely tarnished Deloitte’s reputation. When fake research and mis-quotations make their way into official documents meant to guide public policy, the credibility of not just the firm — but also the decisions based on those reports — becomes suspect.
🔎 Highlighting the Danger of AI “Hallucination” in High-Stakes Work
Generative AI can produce text that looks authoritative — but without proper checks, it can invent references, misattribute statements, and fabricate entire sources. As seen in the Australian and Canadian cases, such hallucinations can undermine entire studies and mislead decision-makers.
🏛️ Implications for Governments and Public Policy
When governments rely on external consultants for key policy decisions — be it health care, welfare, or labour compliance — flawed reports can lead to misguided policies. These scandals raise urgent questions about governance, accountability, and the standards for AI use in public-sector work.
📉 A Warning for the Consulting Industry
As more firms adopt AI tools to improve efficiency, these cases highlight a stark need for rigorous human oversight, transparency about AI usage, and verification of AI-generated content — especially in domains where accuracy is critical.
Responses So Far — Deloitte’s Stance and Growing Backlash
- In both incidents, Deloitte has issued partial corrections: Canada is reviewing and correcting citations; Australia refunded part of the payment and published a revised report.
- Deloitte Canada claims the core findings remain unaffected despite the flawed references.
- But experts and policymakers are not convinced. In Canada, critics say the errors “ undermine confidence in government to do the work necessary to address issues in our healthcare system.”
- Some suggest this could prompt new regulations requiring public-sector consulting firms to disclose when they use AI, and to adopt stricter validation standards for AI-assisted research.
What to Watch — The Fallout and Possible Reforms
- Regulatory push for AI disclosure — Governments may mandate that consulting firms declare when they use AI tools in official reports and document human review protocols.
- Stronger quality-control standards — Consulting firms may adopt stricter editorial and verification processes before delivering AI-assisted reports.
- Industry-wide impact — reputational and financial — Other firms and clients may become wary of AI-assisted reporting, impacting demand or leading to contract clauses banning unchecked AI use.
- Legal and ethical scrutiny — There could be legal ramifications if flawed reports led to policy decisions with negative consequences. Concerns about professional negligence may rise.
Conclusion
The episode where Deloitte fabricated AI-generated research in multiple government-commissioned reports is a wake-up call — both for consulting firms and the governments that hire them. It shows that while AI can be a powerful tool for efficiency, when used carelessly or without due oversight, it can produce dangerously misleading content. As AI use spreads across industries, this scandal underlines a critical truth: human judgment, fact-checking, and professional ethics remain irreplaceable.
