PwC’s Middle East branch has come under pressure following an analysis by GPTZero, the findings of which were verified by the *Financial Times*. In four publications from 2024–2026, concerning, amongst other things, autonomous AI systems, public services and mobility, footnotes were found leading to non-existent web pages, sources that did not corroborate the claims made, and what appears to be a fabricated study on air quality in Riyadh. GPTZero’s tools also indicated extensive use of AI. However, the detector’s result alone does not determine how the texts were prepared. It is primarily the errors in the sources that can be verified.
PwC announced that it would update a “limited number” of citations and emphasised that it has quality control procedures in place. However, the firm did not explain how the errors made it through the publication process. This is significant because thought leadership reports build an adviser’s credibility and help secure contracts.
This is not an isolated incident. EY and KPMG have previously withdrawn reports following similar allegations, whilst Deloitte partially refunded the fee for an erroneous report for the Australian government. At the same time, AI is rapidly becoming the business standard. Stanford HAI reports that in 2025, 88 per cent of organisations surveyed were using it, whilst 70 per cent were using generative AI in at least one area.
A likely consequence will be tighter controls on publications, including mandatory human verification of sources, recording of the tools used, and clear accountability for the final output. Clients may also ask more frequently whether documents were produced with the aid of AI. The regulatory environment is adding to the pressure. From 2 August 2026, the European Union will enforce the AI Act and new transparency rules. For consultancy firms, therefore, what is at stake is not only the quality of an individual report, but trust in the entire business model.
