Four 'thought leadership' reports from PwC Middle East, published between 2024 and 2026, have been found to contain AI hallucinations and fabricated claims, according to an investigation by AI detection software company GPTZero.
One report, 'Transforming Governance' from 2025, was flagged by GPTZero as having an 84% chance of being entirely AI-generated, increasing to 100% when its reference section was excluded. This report promoted a PwC data-driven framework called 'Citizen Pulse' and claimed its use by the governments of Denmark, Saudi Arabia, the United States, and Australia, though GPTZero found no public evidence to support these claims.
GPTZero also noted that citations in this report did not follow standard academic or consultancy conventions, with only 10 of 17 citations corresponding to footnotes. The firm suggested inconsistencies in describing 'Citizen Pulse' as either an existing product or a thought leadership exercise are common symptoms of text generated by a poorly-prompted large language model (LLM).
The other three PwC Middle East reports investigated also contained multiple fabricated claims, including conflating statistics, citing unrelated sources, and asserting false correlations. One report, 'Building a Cyber-Resilient eMobility Ecosystem', included a URL in a citation that referenced ChatGPT.
In response, a spokesperson for PwC Middle East said the firm takes the accuracy of its published research seriously and is updating a limited number of supporting citations. They added that PwC has quality control processes for research and content development that all personnel are expected to follow, consistent with their approach to Responsible AI.
This follows similar findings involving other 'Big Four' firms. Last month, a KPMG report on agentic AI was found to contain AI-generated hallucinations and fake citations. In May, EY withdrew a study on loyalty rewards programmes due to apparent AI hallucinations and fake footnotes. Deloitte also issued a partial refund to the Australian federal government last October after a report contained errors caused by AI use.