PwC reports allegedly contained AI-generated false and fabricated sources
GPTZero says four PwC Middle East reports included fabricated sources and false claims, extending concerns about AI hallucinations across all Big Four firms.
What GPTZero says it found
GPTZero found fabricated sources and false claims in four PwC Middle East reports covering 2024 to 2026, according to The Decoder. The strongest example cited is a governance report titled “Transforming Governance,” which GPTZero rated as 84 percent likely to be entirely AI-generated. The report also promoted a PwC product called “Citizen Pulse” with claims about government use in Denmark, Saudi Arabia, the U.S., and Australia that The Decoder says had no supporting evidence.
Why the citation problem matters
The Decoder reports that the PwC documents with the highest AI scores also had the most false sources. GPTZero describes the pattern as “Vibe Citing,” where references are loosely attached, may include incorrect titles, URLs, or authors, and do not actually support the claims they accompany. For readers and clients, the takeaway is simple: AI-assisted reports still need source-by-source verification before being trusted or shared.
PwC’s response and the Big Four context
PwC Middle East told the Financial Times it takes accuracy seriously and is “updating a limited number of supporting citations,” according to the article. The firm did not explain how the errors entered the reports. GPTZero had previously identified fabricated sources at KPMG, Deloitte, and Ernst & Young, meaning all Big Four firms are now touched by allegations involving AI hallucinations or unsupported citations.