PwC Reports Riddled With AI Hallucinations: GPTZero Investigation

PwC Caught Publishing AI-Hallucinated Content in Middle East Reports

PricewaterhouseCoopers (PwC), one of the world's Big Four accounting and consulting firms, has been found to have published multiple advisory reports riddled with AI hallucinations — fabricated citations, fake footnotes, and unverifiable claims. According to an investigation by AI detection firm GPTZero, verified by the Financial Times, four PwC Middle East reports on artificial intelligence and electric vehicles were largely AI-generated and contained numerous hallucinated sources. The reports were intended to attract new clients in the Middle East, raising serious questions about quality control and responsible AI use at one of the industry's most prominent firms.

What Did GPTZero Find?

GPTZero's investigation analyzed four PwC Middle East 'thought leadership' reports published between 2024 and 2026. The reports covered topics including agentic AI, government services, cybersecurity for electric vehicles, and autonomous mobility. Key findings include:

• Fake footnotes: Numerous citations pointed to non-existent articles, unrelated sources, or webpages that never existed. One footnote even ended with 'chatgpt.com' in its URL.
• Non-chronological footnotes: In one report, footnotes 7–10 appeared on page 7, followed by footnotes 1–6 on page 8 — a nonsensical ordering typical of AI-generated content.
• Hallucinated product: One report, titled 'Transforming Governance: Citizen Pulse,' scored 84% likely AI-generated (rising to 100% excluding the reference section). It promoted a framework called 'Citizen Pulse' that appears to have little public evidence of existing, and falsely claimed that governments of Denmark, Saudi Arabia, the United States, and Australia were using it.
• Misattributed claims: Statements were attributed to wrong individuals, and a study from MIT about AI in government services was cited but could not be verified.

GPTZero described the pattern as showing 'irresponsible AI usage' by the consulting firm. The investigation follows a similar scandal involving KPMG, which withdrew a report on agentic AI in June 2026 after GPTZero found that only 5 of its 45 citations were accurate. The KPMG AI hallucinations scandal highlighted a systemic issue across Big Four firms.

PwC's Response and the Irony Factor

PwC Middle East acknowledged the issues, telling the Financial Times that it takes the investigation 'seriously' and is updating 'a limited number of supporting sources' in the reports. The firm stated that it has 'quality processes for research and content development in line with our approach to responsible AI.' However, critics note the deep irony: PwC has been advising clients on how to adopt AI responsibly, and the same week the scandal broke, PwC US CEO Paul Griggs warned companies about 'bolting AI onto broken processes' — advice his own firm inadvertently proved right.

The incident is particularly damaging for PwC because the firm, like other Big Four members, has positioned itself as a trusted adviser on AI implementation. The Big Four consulting AI ethics debate has intensified as firms rush to monetize generative AI tools while struggling to maintain basic accuracy in their own outputs.

What Are AI Hallucinations?

AI hallucinations occur when large language models (LLMs) generate information that is false, fabricated, or nonsensical but presented as factual. This happens because AI models predict text based on statistical patterns rather than verifying facts against a knowledge base. Common types include:

• Fabricated citations: The AI creates realistic-looking but non-existent academic papers, news articles, or reports.
• Misattributed quotes: Statements are assigned to the wrong person or organization.
• Non-existent entities: The AI invents products, companies, or government programs that do not exist.

GPTZero's Hallucination Check tool specifically targets these issues in professional documents, and has become a key resource for detecting AI-generated slop in high-stakes consulting and accounting work.

Broader Implications for Consulting and AI

The PwC scandal is part of a worrying trend. In June 2026, KPMG withdrew a report after similar findings, and earlier, EY pulled a report on loyalty rewards programs containing AI-generated fake footnotes. Deloitte has also faced scrutiny. These incidents raise fundamental questions about the consulting industry's rush to adopt generative AI without adequate human oversight.

For businesses relying on Big Four advice, the message is clear: AI-generated content can be dangerously unreliable, even when produced by firms that claim expertise in the technology. The responsible AI implementation in consulting will likely become a key regulatory focus as these scandals accumulate.

FAQ

What did GPTZero find in PwC's reports?

GPTZero found that four PwC Middle East reports contained AI-generated text with fabricated citations, non-chronological footnotes, misattributed claims, and at least one hallucinated product (Citizen Pulse). The findings were verified by the Financial Times.

How did PwC respond?

PwC Middle East said it takes the matter seriously and is updating a limited number of supporting sources. It maintains that it has quality processes for research and content development aligned with responsible AI practices.

Why is this incident significant?

It highlights the irony of a firm that advises clients on AI adoption failing to ensure accuracy in its own AI-generated content. It also underscores systemic quality-control issues across Big Four consulting firms as they rush to leverage generative AI.

What are AI hallucinations?

AI hallucinations are false or fabricated information generated by AI models, presented as fact. They include fake citations, misattributed quotes, and invented entities. They occur because AI models prioritize statistical plausibility over factual accuracy.

What can be done to prevent this?

Organizations should implement rigorous human review processes, use AI detection tools like GPTZero, verify all citations against original sources, and establish clear policies for responsible AI use in content generation.