PwC Caught Publishing AI Hallucinations Across Four ‘Thought Leadership’ Reports

On July 29, 2026, AI-detection firm GPTZero published an investigation revealing that four PwC Middle East "thought leadership" reports published between 2024 and 2026 were riddled with AI-generated hallucinations — including fabricated products, fake government deals, and a citation containing "utm_source=chatgpt.com" . The findings make PwC the last of the Big Four consulting firms — following Deloitte, EY, and KPMG — to be caught publishing sloppy AI-generated work while marketing AI expertise to clients . The same day the story broke, PwC US CEO Paul Griggs gave a public interview warning companies not to use AI to paper over broken processes .

What GPTZero's investigation found

GPTZero's team used its Hallucination Check tool and AI Detector to analyze four reports published by PwC Middle East between 2024 and 2026 . The results showed a clear pattern of irresponsible AI usage across all four documents:

"Transforming Governance: Citizen Pulse" (2025) — GPTZero found the report was 84% likely to be entirely AI-generated, with the main body scoring 100% AI-generated . It promoted a PwC methodology called "Citizen Pulse," claiming the governments of Denmark, Saudi Arabia, the United States, and Australia were using it. GPTZero found little public evidence that "Citizen Pulse" exists as a formal PwC product, concluding the firm appeared to have hallucinated "both an entire product and business dealings with four separate nations" .

"Shaping the GCC Mobility Landscape" (2026) — Flagged as 82% Mixed by GPTZero's AI Detector, containing hallucinated citations and unverifiable claims .

"Building a Cyber-Resilient eMobility Ecosystem" (2025) — Scored 60% AI-generated or Mixed, with fake footnotes and fabricated sources .

"Agentic AI – The New Frontier in GenAI" (2024) — Scored 50% AI-generated or Mixed, with hallucinated references .

The reports contained fake footnotes, misattributed claims, made-up sources, and unverifiable statistical data . A particularly revealing detail: one citation included the tracking parameter "utm_source=chatgpt.com," indicating it was copied directly from a ChatGPT output without human verification . GPTZero researchers also found that one report's "real world success story" about J.P. Morgan's use of agentic AI was sourced to a teenage blogger with 280 Medium followers .

PwC becomes the fourth and final Big Four firm caught

GPTZero's investigation against PwC marks the fourth in a series of probes targeting the Big Four. The firm was preceded by:

• Deloitte — GPTZero previously uncovered hallucinated citations in Deloitte Australia reports .
• Ernst & Young (EY) — In May 2026, GPTZero found that 60% of references in a December 2025 EY advisory report on loyalty cybersecurity were hallucinated, including broken links to nonexistent Forbes and WIRED articles .
• KPMG — In June 2026, GPTZero identified hallucinations in KPMG's "Redefining excellence in the age of agentic AI" report. Only 5 of 45 citations matched their sources, and organizations including UBS, the NHS, and Swiss Federal Railways denied the case studies attributed to them. KPMG pulled the report on June 13, 2026 .

PwC is now the fourth and final Big Four firm caught in GPTZero's investigations .

PwC's response

As of the publication date, PwC Middle East had not publicly confirmed the specific hallucination findings. The reports remained accessible online after the investigation was published . The broader PwC network's response has been limited; the firm did not immediately retract or correct the four reports .

Ironic timing of the US CEO's comments

On the same day the GPTZero investigation was published, July 29, 2026, PwC US CEO Paul Griggs gave an interview to Business Insider in which he outlined two major misconceptions companies have about AI .

• First misconception: AI is not fundamentally a technology transformation. Griggs said companies that confine AI work within the CIO or CTO role will continue to "stub their toe" — they should instead treat AI as a "large-scale business transformation" involving "all of your people and all of your processes" .
• Second misconception: AI alone cannot fix a broken business. "If you're layering it on top of messy processes, all AI is going to do for you is tell you how bad your messy process really is," Griggs said .

The irony was widely noted in coverage of the investigation: PwC's top US executive was publicly warning companies against doing precisely what its own Middle East member firm had been caught doing — slapping AI-generated content on top of broken editorial processes without human oversight .

In March 2026, Griggs had also told the Financial Times that PwC partners who fail to embrace AI "are not going to be here that long," signaling a firm-wide mandate to adopt AI-first practices . That same firm now stands accused of using AI so carelessly that it invented entire products and government clients.

Broader implications for the consulting industry

The GPTZero findings against all four Big Four firms raise serious questions about quality control in professional services. The reports in question were "thought leadership" documents designed to drum up consulting work for partners . As one report noted, the pattern of "Vibe Citing" — where references are loosely associated with a topic without being fact-checked — has become endemic across the industry . The Financial Times, which verified GPTZero's findings, reported that the errors were "not subtle" .