AI hallucinations
Coverage of AI hallucinations in the Nexus archive.
- PwC published ‘thought leadership’ reports marred by AI hallucinations
PwC, a Big Four consulting firm, published 'thought leadership' reports that were found to contain AI hallucinations. The firm is being criticized for producing subpar work while promoting its expertise in the technology.
- Conservative activist Robby Starbuck notches major legal win against Google, defamation suit to proceed
Conservative activist Robby Starbuck secured a legal victory against Google as his defamation lawsuit progresses, with a Delaware Superior Court ruling allowing discovery to proceed. The lawsuit alleges Google's AI platforms falsely portrayed him as a 'monster' through AI-generated content, causing emotional distress, and seeks at least $15 million in damages.
- HalluSquatting AI attack could hijack your computer
HalluSquatting is a cybersecurity attack method where AI tools are tricked into downloading malicious software by exploiting their tendency to hallucinate fake repository names. Attackers register these hallucinated names as real repositories to deliver malware, potentially stealing data or recruiting devices into botnets. Researchers from Tel Aviv University, Technion, and Intuit demonstrated the technique against AI coding tools and personal assistants.
- The bench's AI balancing act
Judges are establishing guidelines to regulate AI use in courtrooms while cautiously adopting the technology themselves. Surveys show 60% of judges use AI tools, but concerns about AI hallucinations, privacy, and reliability persist, prompting initiatives like the Judicial AI Consortium to address gaps in training and understanding.
- A report on the benefits of AI was reportedly full of AI hallucinations
KPMG published a paper about the benefits of AI last year. An investigation found that it was full of AI hallucinations.
- An encyclopedia formed from AI hallucinations – what could go wrong?
Halupedia, an AI-generated online encyclopedia, has been discovered by Feedback, featuring entries like the '19nd century' and 'The Society for the Prevention of Unnecessary Tuesdays,' which highlight potential issues with AI hallucinations.
- Top Law Firm Admits to AI ‘Hallucinations’ in Bankruptcy Filing Tied to Alleged Scam Network
Sullivan & Cromwell, a top law firm, admitted to AI-generated 'hallucinations' in a bankruptcy filing for the Prince Group case, leading to fabricated legal citations. The firm stated internal safeguards were bypassed, allowing inaccurate information to be used in the case tied to an alleged scam network.
- AI hallucinations found in high-profile Wall Street law firm filing
Sullivan & Cromwell apologized to a New York federal judge for errors in a major filing caused by AI-generated hallucinations in the Prince Group case. The law firm's co-head of global restructuring, Andrew Dietderich, acknowledged mistakes including inaccurate citations.
- US law firm apologizes after AI hallucinations made it to a legal filing
A US law firm apologized after AI-generated hallucinations appeared in a legal filing. The firm's representative, Andrew Dietderich, acknowledged existing AI policies to prevent errors but admitted procedures were not followed in this case.