EY PULLS LOYALTY REWARDS STUDY OVER AI HALLUCINATIONS

The withdrawal marks another instance of a major firm encountering problems with AI-generated content. Researchers at GPTZero, which specializes in detecting AI-generated text, discovered that EY's study contained false citations and invented data points—common failures when large language models generate plausible-sounding but inaccurate information. EY has not publicly detailed the extent of the errors or how the flawed study progressed through its quality assurance processes before publication. The incident highlights risks as professional services firms increasingly integrate generative AI tools into research and reporting workflows. This follows similar high-profile cases where organizations have released AI-generated content containing factual errors, including false legal precedents and invented company quotes. The EY situation underscores the ongoing challenge of maintaining accuracy when leveraging AI for knowledge work, particularly in client-facing deliverables.