PwC reports riddled with AI-generated errors, fake citations: Probe

PwC is the third Big Four firm to face such scrutiny this year after EY and KPMG came under similar allegations over AI-generated errors in research reports.

PwC has become the latest Big Four firm to come under scrutiny after an investigation found fake citations, misattributed claims and unverifiable information across four thought-leadership reports published by its West Asian arm, raising fresh concerns over consulting firms' use of generative AI in client-facing research.

 

The investigation, conducted by AI detection startup GPTZero and verified by the Financial Times, examined four PwC Middle East reports published between 2024 and 2026, including a playbook for corporate use of agentic AI, a guide for governments on improving public services, and reports on electric and autonomous vehicles and cybersecurity. It found fabricated claims, hallucinated citations and references that either pointed to unrelated webpages, linked to non-existent pages or could not be independently verified.

 

 

PwC is the third Big Four firm to face such scrutiny this year after GPTZero's investigations prompted EY and KPMG to withdraw reports over hallucinated citations, fake footnotes and fabricated claims.

 

What GPTZero found

 

According to the researchers, several citations did not support the claims they were referenced against, while some cited studies and reports could not be traced. The investigation also identified instances of incorrect attributions, inconsistent sourcing, and other tell-tale signs of AI-generated research across the reports.

 

The most serious case, GPTZero said, involved PwC Middle East's 2025 report Transforming Governance, which promoted a framework called "Citizen Pulse". The researchers said they found little public evidence that the framework existed outside the report, despite claims that governments in Denmark, Saudi Arabia, the US and Australia were already using it to improve public services. None of the cited sources supported those assertions, leading GPTZero to conclude that PwC may have effectively hallucinated both the product and its deployment by four governments.

 

“According to our AI Detector, there is an 84 per cent chance that Transforming Governance is entirely AI-generated. If the endmatter (including references) is ignored, the chance increases to 100 per cent.”

 

GPTZero said similar issues were found across the other three reports it examined for the investigation. It found that many citations either did not support the claims they were referenced against or pointed to non-existent webpages.

 

What did PwC say?

 

PwC Middle East was updating "a limited number of supporting citations" in the reports identified by GPTZero, and that it takes the accuracy of its published research, the firm told FT.

 

It added that the firm has quality control processes for research and content development that employees are expected to follow, but did not explain how the errors made it into the published reports.