PwC Middle East Admits Errors in 4 AI-Tainted Reports, Vows Policy Changes
Updated
Updated · TechSpot · Jul 31
PwC Middle East Admits Errors in 4 AI-Tainted Reports, Vows Policy Changes
3 articles · Updated · TechSpot · Jul 31
Summary
At least four PwC Middle East reports published between 2024 and 2026 contained fabricated citations, fake footnotes and other apparent AI hallucinations, prompting the firm to acknowledge mistakes and review its processes.
GPTZero flagged the papers as AI-generated, and multiple outlets including the Financial Times verified examples such as a footnote linking to an unrelated article, broken webpages and a URL carrying the tag "utm_source=chatgpt.com."
One report cited a teenager's Medium post as the sole source for a JPMorgan AI case study, even though the bank project dated to 2017, while an EV report repeated the claim that 90% of road accidents stem from human error with inconsistent or missing sourcing.
PwC Middle East said it takes research accuracy seriously and is updating a limited number of supporting citations, a response that echoes earlier retractions by rivals EY and KPMG after GPTZero flagged their AI-generated reports.