TECH
Anthropic kills Fable, Mythos after US order
Both Ars Technica and TechCrunch report that Anthropic shut down its Fable 5 and Mythos 5 models after a US Commerce Department directive citing national security concerns over a potential jailbreak. Anthropic complied but disagreed, calling the evidence narrow and noting other models have similar capabilities.
PeopleSoft zero-day exploited by ShinyHunters hits hundreds
A critical zero-day vulnerability in Oracle's PeopleSoft software, tracked as CVE-2026-35273 with a 9.8 severity rating, has been exploited by the ransomware group ShinyHunters. The group targeted approximately 100 organizations, primarily in higher education, stealing data and issuing extortion demands. Oracle has provided a stopgap mitigation but no full patch. Mandiant and Rapid7 are offering indicators of compromise.
Meta AI unit employees revolt over soul-crushing work
TechCrunch reports that Meta's three-month-old Applied AI unit is facing employee revolt, with workers describing the environment as a 'soul-crushing gulag' and calling themselves 'draftees.' The discontent follows an internal outburst during a livestream and a petition signed by over 1,600 employees protesting keystroke monitoring. Meta's chief product officer acknowledged the 'brutal' atmosphere, while CEO Mark Zuckerberg admitted mistakes in a memo.
FBI builds replica town to train for cyberattacks
The FBI constructed a 22,000-square-foot replica town on its Huntsville, Alabama campus to simulate real-world cyberattacks. Opened in February 2025, the Kinetic Cyber Range features fully furnished buildings and functioning systems to train investigators. It has trained over 1,400 students from federal and local agencies in ransomware and digital forensics exercises.
German court holds Google liable for AI false statements
The Munich Regional Court preliminarily ruled that Google is liable for false statements generated by its AI Overviews feature, finding that the AI produced independent, new statements not appearing in search results. The court rejected Google's defense that users should verify information, stating that such warnings do not absolve liability. The ruling could have global implications for AI systems that generate new content from multiple sources.
OpenAI faces state attorneys general investigation
A coalition of state attorneys general is investigating OpenAI, issuing a subpoena for documents on user engagement, data handling, and AI models, according to Engadget citing the Wall Street Journal. The company stated it will engage constructively. The probe follows previous state AG scrutiny of AI firms.
Communities block $130B in data center projects in Q1
Ars Technica reports that researchers found 75 data center projects worth $130 billion were blocked or delayed in Q1 2026, the most on record, citing a structural shift in community opposition. The New York Times op-ed highlights that the resistance gives people “a taste of political power.” However, The Atlantic argues the panic is overblown, noting benefits like tax revenue in Loudoun County.
Mistral AI rumored to raise €3 billion at €20 billion valuation
TechCrunch reports that Mistral AI is in early discussions to raise about €3 billion at a valuation of around €20 billion, nearly double its previous valuation from September. The French AI lab has taken a more open approach than US competitors and positions itself as a sovereign European alternative. It has raised significantly less than US rivals OpenAI and Anthropic. Mistral did not comment.
DOJ approves Paramount-Skydance-Warner merger, expanding Ellison media influence
The Department of Justice approved the acquisition of Warner Bros. Discovery by Paramount Skydance, further consolidating the Ellison family's control over American media. Engadget reports that the deal has sparked fear over the future of film, television, and news industries. The DOJ dismissed comparisons to the Disney-Fox merger and labor concerns, while the UK is conducting its own investigation. State attorneys general in California and New York have indicated they will challenge the deal.
Novo Nordisk cyberattack as UK approves Wegovy pill
Pharmaceutical giant Novo Nordisk disclosed a cyberattack that stole pseudonymized clinical trial participant data and exposed healthcare partner information, potentially enabling targeted phishing. The Register notes the attack occurred on the same day the UK approved Novo Nordisk's Wegovy pill as the first daily GLP-1 tablet for weight management.