4 stories in the last 7 days
The latest hallucinations news, distilled by AI into sharp ~100-word summaries. ByteBrief tracks hallucinations across dozens of tech sources and brings you only what matters, updated hourly. Tap any story for the full brief, or open the original source.
GPTZero found only 5 of 45 citations in KPMG's October 2025 agentic AI report matched real sources. The firm coined "vibe citing" for AI-generated false references. Several case studies about UBS, Swiss Federal Railways, and Transport for London were unsupported by cited sources.
KPMG withdrew a report on artificial intelligence adoption after discovering false claims about AI use at UBS and other organizations. The inaccuracies were likely the result of AI hallucinations, according to the Financial Times.

Attorneys on both sides of a legal case simultaneously relied on AI hallucinations, generating fake legal citations and fictitious quotations. This rare but growing problem stems from lawyers improperly using generative AI for legal briefs without double-checking the output. Courts are increasingly frustrated and may impose stiffer penalties.

Lance Eliot examines what happens when an attorney fails to detect that opposing counsel relied on AI hallucinations in court filings. The adversarial system assumes such errors will be caught, but misses raise questions about penalties or sanctions for the failing lawyer.
Summaries by ByteBrief