The research was conducted by AI detection firm GPTZero, which analyzed four PwC reports from the Middle East published between 2024 and 2026. These reports addressed subjects like artificial intelligence, electric vehicles (EVs), self-driving cars, and enhancing government services, as reported by the Financial Times.
According to GPTZero, the reports exhibited fake footnotes, mismatched references, links to non-existent web pages, and unverifiable information. In one instance, a research paper cited in a report could not be located anywhere, raising suspicions that it may have been entirely fabricated by AI.
Furthermore, the researchers discovered that one report had cited a blog post by a teenager on Medium to describe a JPMorgan AI initiative. This raised concerns, as such a significant example should ideally have been validated by a reliable source. They also highlighted that JPMorgan’s AI project dates back to 2017, well before the rise of ChatGPT and other generative AI tools.
GPTZero researcher Paul Esau noted that the reports exhibited traits of AI-generated content. For instance, one report reiterated the assertion that 90% of road accidents stem from human error three times within a span of two pages, albeit with different or no references each time.
“This claim isn’t false, but no human would cite the same fact three times in two pages using three varying sources,” he stated.
Another report asserted that 70% of CEOs in the Middle East believe generative AI will profoundly impact their businesses; however, the footnote directed to a news article that did not reference the survey. Researchers also observed that one reference link contained “utm_source=chatgpt.com” in its URL.
In response to these findings, PwC Middle East stated that it takes erroneous information in its reports seriously and is in the process of rectifying some of the faulty references and citations. “We take the accuracy of our published research seriously and are updating a limited number of supporting citations in the identified reports,” the company informed the FT.
“Consistent with our commitment to responsible AI, we have quality control processes for research and content development that we expect all our personnel to follow,” it added.
