Madison Square Garden security chief John Eversole had staff use facial recognition cameras to follow a trans woman at Knicks games for two years, former staff told Wired. A lawsuit by a former MSG security officer alleges Eversole wanted her kept "away from the players" and got her banned after he "fabricated a stalking allegation."
Vercel said attackers reached some internal systems after an employee's Google Workspace account was compromised through a breach at AI platform Context.ai, then accessed environment variables not marked sensitive and so not encrypted at rest. A small number of customers were affected. ShinyHunters reportedly claimed the hack and offered the data for sale, though its involvement is unconfirmed.
Paris prosecutors summoned Elon Musk and former X CEO Linda Yaccarino for "voluntary interviews" in a cybercrime probe of X covering alleged complicity in spreading child sexual abuse images and sexually explicit deepfakes, which widened after Grok's output. Prosecutors also told U.S. authorities the Grok deepfake controversy may have been orchestrated to boost the value of X and xAI.
Colombian President Gustavo Petro posted on X a fake Noticias Telemundo segment in which anchors report that Ecuadorian crime boss "El Negro Willy" financed President Daniel Noboa's campaign. Telemundo says the clip was altered with AI from a March 23 broadcast that mentioned neither man. The post was shared more than 6,200 times. Petro deleted it after Telemundo sent questions, his communications secretary said.
A researcher says any free Lovable account could read other users' source code, database credentials, AI chat histories and customer data, and that the flaw was reported via HackerOne in March. Lovable first said it "did not suffer a data breach" and blamed unclear documentation. It then admitted a February change had re-enabled access to public projects' chats, saying HackerOne closed the reports without escalation.
Who did it: HYBRID AI Hallucination Cases database
Judge Corey L. Maze of the Northern District of Alabama ordered plaintiffs' counsel in a suit against 3M to show cause why she should not be sanctioned for a brief "riddled with misquotes, false quotes" and apparently fabricated citations. The judge asked whether she made the citations up herself or used an AI platform that did. Her response is due May 1, 2026.
Judge Kai Scott (E.D. Pa.) imposed a $5,000 Rule 11 penalty on lawyer Rajan, who had previously been fined $2,500, after his omnibus motion in Bunce v. Visual Tech. Innovations cited at least one made-up case. The judge found he "haphazard[ly]" submitted an AI-generated document without verifying it, and warned that a third offence would bring referral to the Pennsylvania Disciplinary Board.
Western Australia's AI traffic cameras generated 53,890 infringements between October 8 and April 17, worth more than $29 million, The West Australian reported. Drivers lodged 3,381 review requests, and 2,043 infringements were withdrawn, wiping out more than $1 million. The transport department says an officer reviews every image before a fine is issued.
On his H3 Podcast on April 15, YouTuber Ethan Klein and his lawyer AttorneyTom announced a class action against Nvidia, Meta, Bytedance, Snap, Amazon, OpenAI and Apple. They allege the companies bypassed YouTube's technological protection measures to scrape millions of videos for AI training, in violation of the DMCA's anti-circumvention provisions. They seek statutory damages, injunctive relief and restitution.
Who did it: HYBRID AI Hallucination Cases database
Washington's Court of Appeals, Division 1, affirmed the dismissal of Arthur West's open-meetings suit against the Lower Duwamish Waterway Group. In a footnote, it noted that he provided "fictitious quotations from existent cases" and rebuked the lack of candor. It reminded pro se litigants to check filings for citations "invented or hallucinated by generative artificial intelligence (AI) or any other source."
The National Board for Assessment of Research Misconduct (NPOF) flagged a Lund University doctoral thesis for AI use, saying its reference list contained 14 fabricated references, Universitetsläraren reported in February. The researcher, who keeps the doctoral title, blamed stress and reference tools, and said the aim had been to cite female authors in a male-dominated field.
