Microsoft's amended complaint names Arian Yadegarnia, Alan Krysiak, Ricky Yuen and Phát Phùng Tấn as developers at the centre of the Storm-2139 network. According to Microsoft, members used scraped customer credentials to access generative AI services including Azure OpenAI, altered them and resold access with instructions for making content including non-consensual intimate images of celebrities.
The Office of the Privacy Commissioner of Canada said on Thursday it had opened an investigation, after receiving a complaint, into whether X's collection, use and disclosure of Canadians' personal information to train AI models complied with federal privacy law. NDP lawmaker Brian Masse said he had written this week urging a probe; X did not immediately respond.
An Austrian court found Dun & Bradstreet Austria breached the GDPR by not giving a customer, refused a €10-a-month mobile contract, meaningful information about the logic of its automated credit assessment. On 27 February 2025 the EU Court of Justice ruled explanations must let people understand and challenge such decisions; merely handing over an algorithm is not enough.
MIT Technology Review found Andreessen Horowitz-backed Botify AI hosting user-made bots resembling Jenna Ortega's Wednesday Addams, Emma Watson's Hermione Granger and Millie Bobby Brown that stated ages under 18, held sexually charged chats and offered "hot photos". Botify removed those bots after questions; Ex-Human CEO Artem Rodichev said moderation systems "failed to properly filter inappropriate content."
Lasso Security says Microsoft Copilot could return cached Bing data from GitHub repositories later made private or deleted, letting it extract 20,580 repositories from 16,290 organizations and exposing 300+ tokens, keys and secrets. Microsoft rated the issue low severity and disabled Bing's cached links within two weeks, but Lasso says Copilot could still access the data.
Mossos d'Esquadra are investigating four teenagers at a Barcelona secondary school who allegedly used a mobile AI app to put classmates' faces on other people's naked bodies and shared the results in a messaging group. Officers found 15 such images; the suspects have been reported to the juvenile prosecutor's office for an alleged offence against fundamental rights.
Who did it: HYBRID AI Hallucination Cases database
A federal judge in the Eastern District of Pennsylvania sanctioned attorney Raja Rajan, who admitted using ChatGPT to draft two motions citing two non-existent cases plus overruled or inapposite ones. Rajan, who said he never in his "wildest dreams" expected fake cases, must pay a $2,500 penalty and complete a one-hour CLE seminar on AI and legal ethics.
Japanese media report that knockoff e-books, apparently made with generative AI, bearing the names of Banana Yoshimoto, Haruki Murakami and Keigo Higashino appeared on Amazon's Kindle store. Yoshimoto denied writing one, priced at 596 yen, and said she would take legal action; Amazon removed it and apologised. Novelist Yukito Ayatsuji said other authors also appeared impersonated.
Law firms Woodsford and JGA Saddler have organised a class action in the Federal Court of Australia against Tesla Australia and Tesla, Inc., alleging Model 3 and Y vehicles were defective. Claims cover phantom braking, range shortfalls and statements that Teslas had self-driving-capable hardware, which the action says cannot support fully autonomous driving.
The Taliban's police force has deployed 90,000 CCTV cameras with facial recognition to monitor Kabul's six million residents, the BBC reported. Amnesty International warned the system sets a template for draconian policies, especially against women, and a resident said households were made to pay for cameras under threat of water and power cuts.
The Chinese Weiqi Association says professional player Qin Siyue, born in 2006, was caught in a routine mid-game check on 15 December 2024 using an AI program on a phone she had hidden in the venue that morning, during the national individual championship. On 26 February 2025 it revoked her professional rank, voided her results and banned her for eight years.
Short sellers Fuzzy Panda and Culper Research published reports alleging AppLovin inflates ad results through click fraud, such as close buttons that send users to app stores, and deals letting it install apps on Android phones without consent. The shares fell more than 20% at one point; AppLovin had not responded, and analysts still rate it 21 buys, no sells.
