How PwC Middle East’s AI-Generated Reports Filled a Gap with Fabricated Sources

An investigation by GPTZero and verified by the Financial Times has uncovered that at least four thought leadership reports published by PwC Middle East over the past two years are riddled with AI-generated hallucinations. The reports, designed to burnish the firm’s expertise and attract consulting mandates, contain dozens of footnotes that either link to non-existent webpages or cite sources that do not support the claimed evidence.

Among the most glaring fabrications is an academic paper on air quality in Riyadh that appears to have been entirely invented — no trace of the study exists in the cited journal or among the attributed authors. Another example includes a reference to an obscure teenage blogger with 280 followers on Medium as the sole source for PwC’s description of a JPMorgan initiative it labelled a “real success case” of agentic AI, even though the bank’s work had been public since 2017. One footnote even includes “utm_source=chatgpt.com”, betraying its direct AI origin.

PwC Middle East told the FT it “takes the accuracy of published research very seriously” and is updating a limited number of supporting citations. The firm added that it has quality-control processes for research but did not explain how the errors slipped through. The discovery follows similar retractions forced by earlier GPTZero probes into reports from rival Big Four firms EY and KPMG, which also contained AI hallucinations.

The Reputational Blow to a Firm That Sells Itself as an AI Trusted Advisor

PwC’s Credibility Gap on AI Advisory

The incident cuts especially deep because PwC, like its Big Four peers, has aggressively marketed itself as a guide for companies adopting generative AI — advising on responsible use, risk management, and implementation. The fact that its own Middle East practice produced reports with such sloppy, oversight-free AI output undermines that pitch. Clients may now question whether the firm can be trusted to police AI use in their own organisations if it cannot police its own thought leadership.

The Big Four’s Collective Research Integrity Problem

PwC is not alone. EY and KPMG have already had to retract AI-tainted reports, suggesting an industry-wide challenge. The pressure to produce high volumes of content quickly may be encouraging teams to lean on generative AI without adequate human review. The result is a growing risk that business leaders base decisions on poorly sourced analysis, a threat that compounds as AI-generated material proliferates within professional services.

A Blogger Teen as a Source on JPMorgan’s Automation

The reliance on a little-followed teenager’s blog as evidence of a major bank’s decade-old initiative illustrates how AI-generated citations can fill gaps with irrelevant or low-quality material. The system likely retrieved the blogger’s post as a plausible reference without checking authority or factual basis. For a firm that charges premium fees, citing such a source — and not catching it in review — signals a breakdown in the editorial discipline clients expect.

Steps for Clients to Insist on Research Integrity from Their Consultants

  • Demand a methodology statement. When commissioning or receiving thought leadership, ask consulting firms whether and how generative AI was used in its creation. Request that any AI-assisted reporting includes a documented human verification step for all external citations.
  • Spot-check citations before acting. Before using data from a consulting report to inform strategy, verify a sample of footnotes against their original sources. The PwC case shows that even a basic spot-check could unveil entirely fabricated references.
  • Incorporate research integrity into contracts. Include a clause in consulting engagements that requires independent fact-checking of all externally sourced information, with clear remedies for inaccuracies. This presssures firms to maintain the editorial rigour they publicly espouse.

Risk & Opportunity Assessment

Commercial RiskMediumLoss of client trust could reduce future consulting mandates in the Middle East, though no immediate project cancellations have been reported.
Competitive RiskMediumCompetitors may use PwC’s embarrassment to differentiate themselves on research quality, but the similar incidents at EY and KPMG mean the entire Big Four sector faces scrutiny.
Regulatory RiskLowNo regulatory body has indicated it will investigate, as the errors do not concern audited financial statements.
Reputation RiskHighThe public revelation of fabricated academic papers, a non-existent study, and a link bearing ‘chatgpt.com’ directly contradicts PwC’s self-portrayal as a trusted AI adviser.
Technology DisruptionLowThe technology itself is not threatening PwC’s business model; the risk stems from misuse rather than any displacement effect.
Commercial OpportunityLowWhile some might argue an opening for data-verification or AI-governance services, no immediate new revenue stream arises from the episode.