GPTZero Finds Hallucinations in EY Cybersecurity Report
TL;DR. GPTZero identified widespread AI hallucinations, including fake citations and misattributions, within a 2025 Ernst & Young cybersecurity report. - The report from EY Canada featured fabricated sources and AI-written text, circulating in public information streams. - GPTZero used its Hallucination Check tool to automate detection of such 'vibe citations' in consulting reports. - This incident highlights broader concerns about AI-generated misinformation impacting research and public trust.
- GPTZero's Hallucination Check tool uncovered numerous fake citations and AI-generated content in EY Canada's 'Points of Attack' cybersecurity report.
- The report, credited to three EY employees, included broken URLs, non-existent sources, and misattributed information.
- This investigation is part of GPTZero's broader effort to combat 'vibe citing,' where LLMs accidentally create fake references, impacting data integrity.