A firm that charges clients premium rates to implement responsible AI just got caught doing the opposite. GPTZero, an AI-detection platform, examined four PwC Middle East “thought leadership” reports published between 2024 and 2026 — covering agentic AI, government transformation, and electric vehicles — and found them riddled with hallucinated citations, fabricated sources, and misattributed claims. The Financial Times independently verified the findings. The irony isn’t subtle. It’s structural.
What GPTZero Actually Found
The specific failures read like a blooper reel nobody at PwC bothered to screen before hitting publish.
- A claim that “nearly 70% of Middle East CEOs” see generative AI reshaping their businesses was footnoted to a news article that never mentioned any PwC survey
- A cybersecurity report footnote URL still contained the tracking parameter “utm_source=chatgpt.com” — like posting a screenshot with your notifications visible, except inside a corporate research document
- A JPMorgan AI “success story” was sourced entirely to a Medium blog post by a teenager with roughly 280 followers, describing a project that actually predates ChatGPT by years
- A Riyadh air-quality academic study cited in an EV report cannot be found in any journal, database, or under the stated authors — a source-shaped ghost
One governance report, “Transforming Governance,” scored 84% likelihood of being entirely AI-generated, according to GPTZero’s analysis. That figure climbed to 100% when the reference section was excluded. The same report promoted a PwC framework called “Citizen Pulse,” claiming deployment by the governments of Denmark, Saudi Arabia, the US, and Australia. No public evidence supports any of those claims.
“No human is going to cite the same fact three times in two pages using three different sources.” — Paul Esau, policy analyst, GPTZero. Esau described the pattern as “chaotic signposting” — a telltale sign of AI-generated research stitching references together without coherent editorial discipline.
The Bigger Problem Isn’t PwC
EY and KPMG were already here — both retracted AI-generated reports after GPTZero flagged similar hallucinations, making PwC the third Big Four firm caught in the same pattern.
Three of the largest consultancies on the planet, all selling analytical rigor while quietly letting chatbots do the research. That’s the consulting equivalent of a personal trainer posting fast-food receipts — and charging clients extra for the nutrition advice.
PwC Middle East called these “limited citation issues” and said it would update supporting references, per statements given to multiple outlets. The firm emphasized its existing responsible AI quality controls. Those controls — evidently — failed. Organizations that have been citing Big Four AI reports in board decks or regulatory filings now face a credibility audit of their own. PwC’s hallucinations didn’t stay contained in the document. They may have already traveled further than anyone realized.
