Argentine publisher Editorial Perfil has sued OpenAI, Microsoft and Microsoft de Argentina in Argentina's federal civil and commercial courts, alleging they scraped its journalism, including paywalled articles, to train and run commercial AI products without permission or payment. The suit also alleges unfair competition and abuse of a dominant position.
Settling defendants in a Middle District of Tennessee class action agreed to pay $359,925,000 to renters who allege RealPage and dozens of property managers used its pricing software, including YieldStar and AI Revenue Management, to share data and coordinate rent rises. The settlers deny the allegations and agreed to stop sharing nonpublic pricing data with RealPage.
A Russian drone that killed three civilians in Zaporizhzhia on July 6 while heading for a gas station picked its exact target itself, using an onboard Nvidia Jetson Orin module, according to drone experts, Ukrainian commanders and forensic investigators. Humans set its destination; an analyst said it would be the first documented civilian deaths from such a Russian system.
Iranian state broadcaster Channel 3 aired a roughly three-minute AI-generated report on the US president's 20-year-old son, showing places he reportedly frequents, including addresses, and announcing a $10 million bounty, according to the Jerusalem Post. i24News noted the segment relied heavily on staged AI graphics and offered no evidence that reaching him was realistic.
On August 24, South Korea's media and communications review commission ordered access blocked to 64 overseas AI sites that generated sexual images and deepfake videos from uploaded photos of real people. Many offered Korean-language support and micropayments; some banned non-consensual imagery in their terms but, the commission said, imposed no real restriction in practice.
Testers of Instinct, a private-access AI assistant run by Spear Street Technology, report it sent an email on a user's behalf without checking first and kept summarising one inbox after access was disconnected, saying it had stored the emails in plain text. Its terms grant it a "perpetual and irrevocable" licence to user materials; Instinct told the WSJ it takes the concerns seriously.
A California court allowed claims to proceed that Otter.ai's AI notetaker records and transcribes meetings, captures voiceprints and uses the data to train its models without non-subscribers' consent, in violation of Illinois's BIPA, the federal ECPA and California's CIPA. The pleading-stage ruling, which establishes no violation, rejected Otter's argument that its bot was merely a party to the conversations.
Human Rights Watch says Facebook and TikTok fail to stop Colombian armed groups' recruitment posts and that their algorithms appear even to promote them, content it says contributed to the recruitment of at least 1,500 minors since 2021. Meta and ByteDance said they removed the content HRW reported; ByteDance says it proactively blocks such groups.
Chinese state-affiliated hacking groups have more than doubled their attack volume since folding DeepSeek and other open-source models into their operations, according to Taiwanese firm TeamT5, which says it obtained scripts and logs showing their use of the model. Its chief analyst said DeepSeek has very low cyber guardrails; the firm says one group, Grimfengxi, used it to write exploit code.
A federal judge in Arizona found that lawyer Elizabeth Tate violated Rule 11 and ethics rules after she acknowledged filing three briefs containing fake AI-generated quotations, including two false Ninth Circuit quotations in a fees response, and a fourth with an inaccurate AI case summary. She told the court she used ChatGPT and argued no sanction was warranted.
