AI governance initiative cites Scott M. Graffius research on AI hallucinations

https://scottgraffius.com/blog/files/ai-governance-initiative-uses-graffius-research.html

#AI #AIGovernance #AIHallucinations
#AIResearch

ZDNet: I paid Microsoft’s premium Copilot agents to do my work – they were confidently bad at it. “Over the past few weeks, I’ve been trying to use the AI features in Microsoft 365 and Windows for a variety of everyday work tasks. Copilot shows occasional flashes of competence, but more often, the results I’m seeing are a mishmash of misinformation, hallucinations, and time-wasting dead ends. […]

Mashable: How often does Gemini 3.5 Flash hallucinate or lie? Google isn’t saying.. “As AI gets integrated into every facet of our lives, AI hallucinations remain a stubborn and intractable problem. Yet in the two-hour Google I/O keynote, where Google introduced a massive expansion of AI search and a new default model, Gemini 3.5 Flash, hallucinations didn’t warrant a mention.”

Iowa Capital Dispatch: Attorney reprimanded for AI-fabricated information in court filing. “It’s at least the second time an Iowa attorney has faced allegations of using artificial intelligence, or AI, to author briefs that include false information, often referred to as ‘hallucinations’ since the information is created by AI based on data patterns rather than on verified facts.”

CITP Blog: Can AI reduce burdens on courts by automatically verifying citations?. “Fabricated case citations generated by AI are appearing in court filings at an accelerating rate. Combined with other tracking efforts, we have identified over 1,000 filings containing hallucinated citations from self-represented (pro se) litigants and lawyers alike.”

Microsoft tries reassuring the public that AI is not replacing humanity

https://fed.brid.gy/r/https://nerds.xyz/2026/05/microsoft-ai-not-replacing-humanity/

Hackernoon: When AI Agents Lie to Each Other. “A Datadog State of AI Engineering report from April 2026 found that 1 in 20 AI requests already fail silently in production. The system keeps running. It returns answers that look correct. But the data underneath is wrong. In normal software engineering, a 5% silent failure rate would be a crisis. We would never ship code with those numbers.”

Steven Rosenbaum, the author of a nonfiction book about the effects of AI on truth, acknowledged yesterday that his book includes numerous made-up or misattributed quotes concocted by AI.

The fake statements found in Rosenbaum's The Future of Truth: How AI Reshapes Reality, were discovered and brought to light by The New York Times, which itself has been caught in the past using AI hallucinations in its reporting.

Most hilarious part of all: Rosenbaum acknowledged the errors but noted that they highlight the exact premise of his book: that AI-assisted research has severe flaws and that verification remains a major societal risk 😂 🤦🏽‍♀️

#AI #Books #AIhallucinations
https://www.nytimes.com/2026/05/19/business/media/future-of-truth-ai-quotes.html