The parents of a 31-year-old woman with severe epilepsy sued OpenAI, alleging ChatGPT practiced medicine without a license and framed her doctors "as obstacles to manage." Chat logs seen by The New York Times show it drafting plans to cut her seizure drug from 400 milligrams to zero before her death in February; OpenAI says the GPT-4o version involved is no longer available.
The Independent Chief Inspector of Borders and Immigration is considering investigating Home Office asylum tools built on OpenAI's GPT-4, WIRED reports. Open Rights Group found the share of AI case summaries deemed unfit for use significantly higher than the published 9 percent error rate; the Home Office says the tools do not make decisions.
Privacy group noyb filed five complaints against Scale AI in France, Italy, the Netherlands and Austria, alleging its Outlier platform illegally monitors freelancers who produce AI training data, with up to 19 trackers recording keystrokes and mouse movements and repeated biometric checks via Persona. It asks regulators to fine Scale; the company did not immediately respond.
Ten Norwegian media groups, including NRK, Schibsted and Amedia, asked the Oslo District Court for a provisional injunction against Meta's new commercial terms, which they say would let it collect their articles, photos and videos, paywalled ones included, to develop AI products. Meta has postponed the rollout to 20 November after pushback.
A lawsuit filed Sept. 9 in Buncombe County Superior Court alleges Asheville Christian Academy failed to stop students from bullying and hazing a student, including by creating AI-generated sexually explicit imagery of him and sending his personal information to a known pedophile. The private Christian school denied the allegations on Oct. 27.
William Hamish Yeates, 20, the first person charged under Australia's 2024 federal deepfake pornography laws, was sentenced to two years and three months but released on a three-year good-behaviour order. He admitted creating and sharing sexually explicit deepfakes of a teenage girl; the judge found no "malicious motive" and said his youth and autism reduced his culpability to some extent.
South Africa's Labour Court halted an arbitration after finding that a CCMA commissioner allowed six dismissed Post Office employees' claims to proceed by citing a Constitutional Court judgment that does not exist. Acting Judge Sean Snyman said in his October 23 ruling the commissioner "must have relied on AI" and called it "a fundamental failure of justice."
Bots submitted 90,835 of the 118,756 responses to the UK's 2023 consultation on the tobacco and vapes bill, nearly all opposing the disposable vape ban, according to research published in Tobacco Control. The health department says its controls caught and excluded them, and the ban became law anyway.
British Columbia's privacy commissioner has been investigating since June 10 how OpenAI handled personal information the Tumbler Ridge shooter gave ChatGPT, a probe revealed only through OpenAI's court petition to quash it. OpenAI calls the commissioner's document order overbroad and says it already gave the shooter's chat logs to the RCMP; the watchdog confirmed the probe on Oct. 27.
