PwC is the latest Big Four consulting firm whose published reports have been found to contain apparent AI hallucinations and problematic citations.
PwC Middle East is facing scrutiny after researchers at GPTZero found AI hallucinations, fabricated citations, and unverifiable claims across four of its public "thought leadership" reports, adding to a growing list of similar issues involving the Big Four consulting firms.
The findings, first verified by the Financial Times, center on reports published between 2024 and 2026 covering topics including agentic AI, government services, cybersecurity for electric vehicles, and autonomous mobility. According to GPTZero, the reports contained fake footnotes, citations that pointed to unrelated or nonexistent sources, and claims that could not be verified.
Investigation highlights fabricated sources
GPTZero said one of the reports, Transforming Governance, showed the strongest signs of AI-generated writing. The researchers estimated there was an 84% chance the report was entirely AI-generated, rising to 100% when its reference section was excluded.
Among the issues identified were references to a Riyadh air quality study that investigators could not find in the cited journal or under the listed authors, footnotes linking to webpages that did not support the accompanying claims, and citations pointing to pages that no longer existed or never existed.
Researchers also questioned claims that PwC's "Citizen Pulse" framework was already being used by the governments of Denmark, Saudi Arabia, the United States, and Australia, saying they found no public evidence supporting those deployments.
Another report referenced a JPMorgan automation initiative using a Medium post from a teenage blogger as supporting evidence, while one cybersecurity report included a URL containing the parameter "utm_source=chatgpt.com," which GPTZero cited as another indicator of AI-assisted drafting.
PwC says it is updating citations
Responding to the findings, PwC Middle East said it is correcting the affected reports.
"PwC Middle East takes the accuracy of our published research seriously and is updating a limited number of supporting citations in the reports mentioned. Consistent with our approach to Responsible AI, we have quality control processes for research and content development we expect all our people to adhere to," a PwC Middle East spokesperson said, per City AM.
The firm did not explain how the reported errors made it into the published documents. Paul Esau, a researcher at GPTZero, told the Financial Times that inconsistent citation patterns were one clue that AI had likely been used.
"This claim isn’t fake, but no human is going to cite the same fact three times in two pages using three different sources," Esau said, referring to repeated citations about human error causing 90% of traffic accidents.
AI errors spread across the Big Four
PwC is the latest Big Four consulting firm to face scrutiny over suspected AI-generated research and unreliable citations.
Earlier GPTZero investigations led EY and KPMG to withdraw reports that allegedly contained fabricated or unverifiable references. Deloitte separately issued a partial refund to the Australian government after AI-related errors were identified in a report it produced under a government contract.
The PwC case carries an added reputational risk because some of the affected reports were intended to demonstrate the firm’s expertise in AI and emerging technologies — areas in which it advises paying clients.
What this means for businesses
The episode underscores a growing challenge as organizations increasingly rely on generative AI to produce research and marketing materials.
AI can dramatically speed up drafting, but the PwC case shows that unchecked citations and fabricated references can undermine credibility, especially for firms advising clients on responsible AI adoption. For businesses, the takeaway is that AI-generated content may still require rigorous human verification before publication, particularly when reports are intended to guide strategic or policy decisions.
Also read: EU Sets Rules Roadmap for Deepfake, AI Content Labels