TECH
UK to ban social media for under‑16s, upcoming restrictions detailed
UK Prime Minister Keir Starmer announced a ban on social‑media use for children under 16, with legislation expected before year‑end and possible rollout by spring 2027. The Verge notes the ban will cover major platforms such as TikTok, Instagram, Facebook, Snapchat and X, while excluding messaging apps, and will also block stranger chat in games, livestreams and romantic chatbots. TechCrunch adds that the scope may extend to Reddit, Threads, Twitch and Kick, and highlights privacy concerns and the need for new legislation.
Anthropic disables Fable and Mythos models after US trade directive
Anthropic announced it has abruptly disabled access to its Fable 5 and Mythos 5 AI models following a verbal directive from the U.S. Commerce Department that places the models under export controls. The government cited a reported “jailbreak” that could let the models analyze code for security vulnerabilities, prompting concerns about national‑security risks. Ars Technica highlighted the directive and the alleged jailbreak, while other outlets have noted that Amazon also raised security concerns about the models.
FISA Section 702 expires but surveillance will persist through 2027
Ars Technica reports that Section 702 of the Foreign Intelligence Surveillance Act officially sunset at midnight, but existing year‑long certifications issued by the FISA Court keep the program active until at least March 2027. The outlet notes that the Brennan Center and Cato Institute stress the lapse does not halt data collection, and the Electronic Frontier Foundation warns the government could also rely on Executive Order 12333. Rep. Jamie Raskin (D‑Md.) said surveillance will continue unchanged while lawmakers remain divided over reforms.
SpaceX goes public with $1.8 trillion valuation focused on AI services
Ars Technica reports that SpaceX began trading on Nasdaq at $135 per share, quickly climbing to $160.95 and valuing the company at about $1.8 trillion, making Elon Musk a trillion‑plus billionaire. The filing emphasizes that the bulk of the company’s future profit is expected to come from AI compute services delivered from orbital data centers, rather than its traditional NASA contracts or Starlink business. Investors are now watching how SpaceX will allocate resources between lucrative AI contracts and its ongoing space missions such as Artemis.
ShinyHunters exploits PeopleSoft zero‑day, steals gigabytes from hundreds
Researchers at Mandiant and Google’s Mandiant team say the SSRF vulnerability (CVE‑2026‑35273) in Oracle’s PeopleSoft, rated 9.8, was exploited by the ShinyHunters ransomware group for over two weeks before Oracle issued a mitigation. The attackers accessed about 300 endpoints across roughly 100 organizations, mostly higher‑education institutions, stole gigabytes of data and posted up to 48 GB from a single victim on their public data‑leak site, demanding ransom. Oracle has released a stop‑gap fix but not a full patch, while security firms Rapid7 and Mandiant have published indicators of compromise and remediation steps for PeopleSoft users.
Meta tests Pentagon‑linked face‑recognition software in smart glasses
According to WIRED, Meta licensed facial‑recognition technology from Rank One Computing, a contractor whose customers include U.S. law‑enforcement and military agencies, to prototype the feature for its Ray‑Ban and Oakley smart‑glasses app. The code was embedded in a test version of the Meta AI app shipped to millions of phones but remained dormant and was removed after the report. Both Meta and Rank One declined to comment on the partnership.
AI layoffs surge while hiring spikes as firms seek oversight staff
Tech firms are cutting tens of thousands of jobs, often blaming AI, even as they post strong earnings. At the same time, a growing cadre of AI oversight staff is being recruited, inflating the fortunes of a small insider class. The contrast highlights a widening gap between laid‑off workers and rapidly wealth‑creating AI ventures.
Ukraine tested fully autonomous drones that killed Russian soldiers, officials say
Ars Technica reports that a Ukrainian drone maker disclosed a one‑time battlefield test in which fully autonomous quadcopter drones, operating in a “Terminator mode,” killed a handful of Russian soldiers. Ukrainian officials quoted by New Scientist say the government prohibits AI from making final kill decisions and that combat drones are only semi‑autonomous, with human operators retaining control. A CSIS analysis cited in the same coverage notes that while fully lethal autonomous weapons remain rare, Ukraine is increasingly fielding drones with autonomous navigation and target‑recognition functions to offset electronic‑warfare challenges.
NASA's X‑59 reaches Mach 1.4 and 55,000‑foot altitude
Engadget reports that NASA's X‑59 research plane hit Mach 1.4 and climbed to 55,000 feet, surpassing a prior Mach 1.1 test. The achievement clears a critical step toward the Quesst mission, where the aircraft will fly over populated areas to gauge public reaction to its quiet "sonic thump." The team will now conduct acoustic validation to confirm the low‑boom signature before community flights.
Fox to acquire Roku in $22 billion deal, creating third‑largest US streaming player
The Verge reported that Fox announced a $22 billion acquisition of Roku, merging its TV networks and Tubi streamer with Roku's devices, smart‑TV software, and The Roku Channel. The combined company aims to become the third‑largest television player in the United States by viewing share while keeping Roku an open platform. The deal, which includes Roku founder Anthony Wood joining Fox's board, is still subject to regulatory approval.