PwC becomes the latest Big Four consulting firm to face criticism over the careless use of generative AI. An investigation found that several thought leadership reports by PwC Middle East contained fabricated citations, broken hyperlinks, unverifiable claims and references to academic papers that do not exist. The findings have raised fresh questions about editorial standards and AI governance within global consulting firms.
Researchers from GPTZero found that four PwC publications were using AI. These included reports on enterprise adoption of AI, autonomous agents, government services, and trends in electric vehicles. Inaccuracies included fake footnotes, wrong source attributions, and links that either did not lead anywhere or were irrelevant to the information presented. The Financial Times confirmed these results.
The most striking one was the citation URL that featured "utm_source=chatgpt.com", suggesting the use of unverified content generated by AI. Also, researchers found mentions of non-existing research papers and unconfirmed statistics. These findings only proved once again that AI-generated material had not gone through a proper human review process before publication.
The incident gained spotlight as PwC actively advises governments and enterprises on responsible AI adoption and governance. Following the revelations, the firm said, “We are updating a limited number of citations and remain committed to maintaining the accuracy of our research.” However, it has not explained how the flawed reports passed internal quality checks.
PwC is not alone; previous research by GPTZero found that other Big Four companies such as EY and KPMG had withdrawn their reports following the discovery of similar AI hallucinations. The repeated errors point out that there might be challenges for consulting firms in trying to balance between fast production of content using AI and verification through editing.
Also Read: Abu Dhabi Increases Investments in India's Key Industries