In the modern corporate world, the promise of artificial intelligence has often felt like a magic bullet—a way to synthesize massive amounts of data, draft reports in record time, and solidify a firm’s position as a thought leader. However, a recent incident involving the global consulting giant EY serves as a sobering reminder that speed and automation are poor substitutes for human oversight. The firm recently took the embarrassing step of retracting a study focused on loyalty reward programs after independent researchers discovered it was riddled with “hallucinations”—a polite tech term for fabricated data and entirely imaginary sources. For a firm that prides itself on precision and professional integrity, the presence of fake statistics and nonexistent footnotes is more than just a glitch; it is a reputational blow that highlights the risks of rushing to adopt AI without verifying the output against reality.
The errors in the EY Canada report, titled “Points of Attack: Uncovering Cyber Threats and Fraud in Loyalty Systems,” were not subtle nuances that required a forensic accountant to uncover. Researchers from the detection platform GPTZero found that the document suffered from catastrophic logical inconsistencies, such as claiming the loyalty scheme market was worth $200 billion while simultaneously claiming, in the same breath, that unclaimed points amounted to that exact same figure. Even more damaging was the discovery that major sections of the piece were supported by references to a nonexistent McKinsey report and a collection of footnotes that led to dead URLs or completely irrelevant web pages. By relying on an AI model to generate the heavy lifting of the research, the consultants involved failed to perform the most basic duty of their profession: reading what they were actually publishing.
The implications of this incident stretch far beyond one faulty report. As the researchers at GPTZero pointed out, the internet acts as a shared pool of knowledge; when high-profile firms like EY “poison the well” by publishing AI-generated misinformation, that data propagates, eventually contaminating future research and public discourse. It creates a perverse cycle where AI models of the future digest today’s fabricated reports to create tomorrow’s falsehoods. When a respected brand with a massive online footprint hosts flawed information, it lends a veneer of authority to lies that might otherwise be easily debunked. This underscores a dangerous paradox: firms are eager to be seen as the vanguard of AI innovation, yet they are increasingly vulnerable to the very systems they are marketing to their own clients as transformative tools.
Unfortunately, EY is far from being an outlier in this trend. The professional services landscape has become a minefield of AI-induced errors. Last year, Deloitte faced a similar predicament when a report prepared for a Canadian provincial government was found to contain completely fabricated academic citations. Similarly, the prestigious law firm Sullivan & Cromwell recently found itself apologizing to a New York court after filing documents that quoted non-existent bankruptcy laws and invented legal precedents. These are not merely cases of “teething problems” with new software; they represent a fundamental erosion of the “due diligence” standard that professional services firms are paid millions of dollars to uphold. There is a palpable tension between the pressure to sound cutting-edge and the duty to be accurate.
Inside boardrooms, the rhetoric surrounding AI is almost exclusively optimistic. EY, for instance, has boasted that its AI-related revenue grew by 30 percent in the last year, with thousands of staff members supposedly trained to implement these tools for client projects. Firms are pouring billions into these technologies, convinced that if they don’t lead the AI revolution, they will be left behind. Yet, the rush to scale and the desire to “do more with less” appears to be outstripping the internal controls required to manage such powerful, unpredictable tools. When consultants use AI to draft marketing pamphlets or strategic white papers, they are essentially outsourcing their credibility. If an AI can hallucinate an entire report for a public website, one has to wonder about the quality of the “governance frameworks” these firms are building for their clients in the real world.
Ultimately, EY’s apology and the removal of the suspect report from their website are necessary first steps, but they do little to quiet the growing skepticism regarding the corporate adoption of generative AI. The firm maintains that it is reviewing the internal lapses that allowed such a document to slide through their quality control process and has emphasized that the report was not commissioned for a client. While that may provide some comfort to their current partners, the incident has highlighted a massive cultural shift that firms must address immediately: the loss of human skepticism. In the race to automate, these companies must not forget that intelligence—true professional intelligence—requires the ability to discern fact from fiction. Without a return to rigorous fact-checking and human accountability, AI will continue to act less like a sophisticated consultant and more like an eloquent, highly confident liar.