PwC, one of the “Big Four” consulting firms, has been caught red-handed publishing “thought leadership” reports about AI riddled with lazy and easily-spotted hallucinations.
As the Financial Times reports, four reports published between 2024 and 2026, were meant to solicit consulting work for partners in the Middle East. But as investigators from AI detector firm GPTZero found, the reports were an utter mess, chock full of “hallucinated (vibe) citations, fabricated claims, and incomprehensible drafting and formatting decisions.”
The offending reports contained “claims that are unrelated to or contradicted by the cited source.” For instance, a 2025 report included an entire section about a framework known as “Citizen Pulse,” which it alleged was being deployed by governments around the world. However, GPTZero found that the framework didn’t actually exist and was entirely hallucinated by an AI, including supporting citations.
The findings highlight just how pervasive AI hallucinations have become, from lawyers being reprimanded by judges after being caught including fake legal citations in court documents to entire academic fields being inundated with dubious AI-generated papers.
Even prestigious consulting firms are providing their clients with shoddy and entirely unverified slop, raising plenty of uncomfortable questions about the quality of their professional services.
Just last month, another Big Four firm, KPMG, was caught including copious amounts of AI hallucinations in a report about how to integrate AI agents “across investment advisory, risk management and compliance monitoring.”
The latest case is particularly egregious considering several of PwC’s slop-heavy reports were explicitly about “agentic” AI bots, including a guide for governments on how to improve public services for autonomous vehicles.
The extent of the slop is baffling. One of the PwC reports included a footnote referencing a teenage blogger on Medium with 280 followers being cited as a source for a JPMorgan agentic AI initiative.
GPTZero researcher Paul Esau found that a different PwC report had cited the same (correct) claim that human error is responsible for 90 percent of traffic accidents three times in just two pages, while using three different sources, something that “no human” would do, as he told the FT.
In response, PwC downplayed the findings and failed to take accountability.
In a statement to the newspaper, PwC Middle East claimed that it “takes the accuracy of our published research seriously and is updating a limited number of supporting citations” in the reports. “Consistent with our approach to responsible AI, we have quality control processes for research and content development we expect all our people to adhere to.”
More on hallucinations: Consulting Firm’s Report on How Awesome AI Is Found to Contain Idiotic AI Hallucinations