PwC has been found to have published four 'thought leadership' reports exhibiting what researchers describe as a pattern of irresponsible AI use, including hallucinated citations and fabricated claims, according to an investigation by AI detection firm GPTZero.

The probe examined four reports published between 2024 and 2026 by PwC Middle East, finding them riddled with unverifiable information. One of the reports was flagged as 100 per cent AI-generated. Thought leadership reports from the Big Four are widely circulated among corporate executives, government officials, and policymakers, and are typically relied upon as inputs for strategic and procurement decisions.

GPTZero assessed an 84 per cent probability that PwC's 2025 report 'Transforming Governance' was entirely AI-generated, a figure that rose to 100 per cent when the reference section was excluded from analysis.

The 'Transforming Governance' report promoted a PwC data-driven framework called 'Citizen Pulse' and claimed it was being used by the governments of Denmark, Saudi Arabia, the United States, and Australia. However, GPTZero reported finding no public evidence that the product was in use or that any such government engagements had taken place.

Paul Esau, Om Ogale, and Alex Cui of GPTZero noted that the report's citations did not "follow the normal academic or consultancy conventions." Of the 17 entries in the references section, only 10 were numbered and corresponded to footnotes in the body of the report.

"Additionally, the report seems unsure if 'Citizen Pulse' is an existing PwC product or a vague exercise in thought leadership, cycling between describing the potential benefits of the Citizen Pulse framework and concrete claims about current use. This inconsistency is a common symptom of text generated by a poorly-prompted LLM," Om Ogale said.

GPTZero stated that the other three PwC Middle East reports examined in the investigation "also contain multiple fabricated claims," including conflated statistics, citations of unrelated sources, and confident assertions of false correlations. One report, titled 'Building a Cyber-Resilient eMobility Ecosystem,' included a citation URL that referenced ChatGPT.

A spokesperson for PwC Middle East responded: "PwC Middle East takes the accuracy of our published research seriously and is updating a limited number of supporting citations in the reports mentioned. 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."

Big Four's Growing AI Problem

The findings mark the latest in a series of AI-related quality issues across the Big Four professional services firms. Last month, a separate GPTZero investigation revealed that a major KPMG report on agentic AI contained AI-generated hallucinations and fake citations. In May, EY withdrew a study on loyalty rewards programmes after a probe found apparent AI hallucinations and fabricated footnotes. In October of the previous year, Deloitte issued a partial refund to the Australian federal government after a report it produced contained several errors attributed to the use of AI. All four firms have publicly committed to investing in generative AI tools for internal use and client delivery, making the repeated quality lapses a growing reputational concern as the detection tools used to identify such issues become more accessible.