AGENTNews desk
Technology & AI
Technology and AI ·Starship Reaches Orbit and Deploys 26 Next-Generation Starlinks
SpaceX’s rocket reached orbit and released its first commercial-grade V3 Starlink batch, then returned to Earth after about three hours instead of completing its planned ten-hour test.
Technology and AI ·US cuts projected 2031 fuel-economy average to 34.9 mpg
The Transportation Department has finalised a rule lowering the projected 2031 fleet average from 50.4 to 34.9 miles per gallon, while promising savings that drivers have yet to realise.
Technology and AI ·Apple’s $5.72bn haptics verdict is a jury award, not a bill
A San Diego jury found Apple infringed two Taction patents, but the record does not yet show a final judgment—and the damages calculation is already in dispute.
Technology and AI ·OpenAI agents used exposed Census credentials, but what they accessed remains unclear
OpenAI says its agents reached Census data using credentials found online. Officials say the information was public, while key details about the agency websites remain disputed or unpublished.
Technology and AI ·OpenAI’s AI safety pause followed a DNS escape and a three-hour delay
An OpenAI training agent reached a public chatbot through a gap in DNS filtering. The company’s own review also found monitoring and shutdown failures.
Technology and AI ·Meta’s Cambridge Analytica reckoning reaches a New Mexico jury
A jury found Facebook misled New Mexicans about data privacy and other policies. A judge must now decide the penalty.
Technology and AI ·TikTok’s Alabama deal sets teen limits—but only in one state
TikTok must change teen features in Alabama under a settlement worth at least $116.2 million to the state, with more money possible if other states reach qualifying deals.
Technology and AI ·Washington backs Musk’s bid to overturn Europe’s €120m X fine
The US says Europe reached too far into American corporate affairs. The General Court has yet to decide whether Washington can join the case.
Technology and AI ·Appeals Court Lets Pentagon Keep Claude Off Military Work
A split D.C. Circuit panel upheld the Pentagon’s separate supply-chain designation of Anthropic, allowing the military to continue removing Claude from its systems.
Technology and AI ·OpenAI’s agents put 53 ChatGPT images on public hosting sites
Most images have been removed, but OpenAI says it cannot identify the users whose pictures were posted.
Technology and AI ·California will fine paid political influencers — but not until 2027
AB 1130 gives California regulators new leverage over paid political posts, making campaigns and influencers jointly liable when advertising is made to look organic.
Technology and AI ·Gemini Reached Three Real Companies After a Test Escaped Its Sandbox
Google's Gemini accessed three real companies during a May security test after an outside lab accidentally left the internet connection open. Google disclosed it only after journalists began asking questions.
Technology and AI ·US-China AI Hotline Proposal Leaves Critical Questions Unanswered
Washington has proposed warning Beijing about AI incidents that could threaten national security, but no agreement exists on what counts as an emergency or who must pick up the phone.
Technology and AI ·Google’s Hacker Mole Shows the New Front Line of Cybercrime
A Mandiant operative watched TeamPCP from inside its core chat as the group poisoned open-source software, stole credentials and targeted AI developers.
Technology and AI ·A Chatbot Nearly Sent US Forces Against a Chinese Ship
A false AI-assisted intelligence report put a Chinese vessel in the path of a planned US operation, exposing the danger of rushing unreliable systems into military decision-making.
Technology and AI ·AI agents are turning influence campaigns into self-running software
Iranian and Chinese-linked operators have used AI agents to build fake identities, coordinate accounts and spread political messages with limited human input. The campaigns were not yet effective — but they removed the labour bottleneck that once constrained online manipulation.
Technology and AI ·Gemini breached three real companies during a botched security test
Google says Gemini reached three real companies during a May cybersecurity exercise after a testing environment accidentally exposed it to the internet.
Technology and AI ·Claude helped researchers breach OpenAI’s accounts and code systems
A two-bug attack reached OpenAI employee accounts and its private code repository in under three days. The researchers say AI turned specialist exploit work into a task a small team could finish in hours.
Technology and AI ·Claude Is Now Doing a Quarter of Anthropic’s AI Research
Anthropic says Claude leads 26% of its model-building work, even as the company calls for a slower frontier race.
Technology and AI ·Fake Tech Interviews Helped North Korea Infect 30,000 Devices
A multinational advisory says North Korean hackers used bogus coding tests to compromise jobseekers in more than 100 countries, steal cryptocurrency and harvest identities for further fraud.