Technology
- Tue. 11/03
Google says it shut down XCPE, a botnet of 1.3 million hijacked TV streaming boxes
The boxes shipped with hidden software that let criminals use them for ad fraud and to route strangers' internet traffic. Here's how to check your own device.
- Tue. 11/03
CrowdStrike to buy Palo Alto's Fiddler AI and Dreadnode to watch and test AI agents
The cybersecurity company wants tools that record what AI agents do and attack them before criminals can. The price was not disclosed.
- Tue. 11/03
UK cyber agency warns AI-powered ransomware will likely hit Britain's power, water or food systems by 2028
The head of the National Cyber Security Centre says AI isn't inventing new attacks, it's finding old, unfixed weaknesses faster, and he is urging companies to fix them now.
- Tue. 11/03
Stanford launches five-year project to build AI models of the human brain
The Digital Brain Project will scan volunteers' brains with two advanced imaging machines and train AI on the data, hoping to test treatments for depression or Parkinson's on a computer first.
- Tue. 11/03
China's Meituan releases LongCat-2.5, a giant AI model trained on Chinese-made chips
The food-delivery company's model has 1.6 trillion parameters but uses only about 48 billion for each word, and it is built to work through long coding jobs on its own.
- Tue. 11/03
Mountain View's Camect links its AI security-camera alerts to police dispatch centers
A partnership with 3Si lets security staff who confirm a crime on Camect-connected cameras send the video and location straight to local police dispatchers.
- Tue. 11/03
Could special chips pause AI's race? A UC Berkeley researcher weighs a 'hardwired' stop
The idea: computer chips that can run today's AI models but can't train more powerful ones. Andrew Reddie says it's promising but faces big technical and political hurdles.
- Tue. 11/03
South Korea opens a "data factory" where humanoid robots practice chores
A state research institute will have 12 remote-controlled robots record kitchen, laundry and car tasks to train robot AI.
- Tue. 11/03
OpenAI quadruples rewards for hackers who find flaws in ChatGPT
The San Francisco company will pay up to $172,000 for the worst bugs and is bringing in an outside firm to attack its own systems with help from its AI.
- Tue. 11/03
Google brings its AI daily-story app Dreambeans to India, its first market outside the U.S.
The experimental Google Labs app builds illustrated story collections from a user's own Gmail, Calendar, Photos and YouTube, now in 14 languages.
- Mon. 11/02
Korean researchers teach a humanoid robot to walk and handle objects in one sequence without full demonstrations
Seoul National University's method lets a robot learn walking and grasping separately, then chain them together, a step toward robots that can fetch and carry.
- Mon. 11/02
OpenAI will shut down the old ChatGPT desktop app on Jan. 30 and plans to merge its Chat and Work tabs
People still using "ChatGPT Classic" on their computers will need to switch to the newer app, which keeps their chat history.
- Mon. 11/02
MIT scientists who used AI to find new antibiotics win the Pritzker Prize for AI in science
Regina Barzilay and James Collins helped discover halicin, the first antibiotic found by artificial intelligence, as drug-resistant germs spread.
- Mon. 11/02
Microsoft AI chief Mustafa Suleyman says companies should not try to build conscious AI
In a weekend essay on X, he said today's AI is not conscious and that he would call for a halt to AI development if signs of consciousness appeared.
- Mon. 11/02
Robots race up thousands of stone steps on China's Mount Tai in a new climbing contest
Nearly 90 teams sent walking humanoids, four-legged robot dogs and wearable exoskeletons up a famous mountain trail in a test of real-world balance.
- Mon. 11/02
Anthropic says it may soon reach its own warning line for AI that speeds up AI research
The San Francisco maker of Claude added the warning to its updated stock-sale filing Monday, while saying none of its current models has crossed the line.
- Mon. 11/02
Hundreds of harmless-looking AI apps secretly made sexual deepfakes, Wired investigation finds
Pet-portrait and tutoring apps passed Apple and Google reviews, then switched on hidden abusive image tools, researchers say.
- Mon. 11/02
MIT researchers say an AI fixed another AI by editing its 'brain' directly, without retraining
In an unreviewed study, OpenAI's GPT-6 Astra changed 0.016% of a small model's settings and its math scores jumped from 8% to 48%.
- Mon. 11/02
New global standard lets warehouse robots from different makers share the same floor
ISO 21423 sets a common language so mobile robots can tell each other where they are and where they're going.
- Mon. 11/02
Apple is building a $1,000 tabletop robot with a swiveling screen, report says
The device, possibly called HomeView Ultra, could arrive in 2027 or slip to 2028, according to The Information.
- Mon. 11/02
Anthropic tells its Claude coding AI to stop copying other people's video games
A two-sentence rule added to Claude Code aims to block copyright problems, but some developers say it also blocks legitimate work.
- Mon. 11/02
New York Times sues San Francisco’s Brave, saying it copies and sells Times articles to AI companies
The suit also targets a Brave browser setting that the Times says gets around its paywall.
- Mon. 11/02
San Jose’s MetaGuard AI launches AI ‘attackers’ and ‘defenders’ that drill small businesses’ cyber defenses
The Quantum Ventura spin-off prices its new simulation tool from $4,999 a year and a code checker from $3,999.
- Mon. 11/02
Musk says Grok Bot’s costs were ‘unsustainably high’ after users’ weekly limits suddenly ran out
Some subscribers to SpaceX’s AI agent found their allowances maxed out over Halloween weekend.
- Mon. 11/02
Pleasanton’s 10x Genomics launches an AI agent that analyzes lab data from plain-English questions
Sentira Single Cell, built on Anthropic’s Claude, aims to let biologists skip writing analysis code.
- Mon. 11/02
New AI test of expert-level work finds even the best model passes only about 12% of tasks
Research firm SemiAnalysis built GDPx from real tasks in fields such as chip design, biology and finance. Models scored near zero in finance and law under its strict pass-or-fail grading.
- Mon. 11/02
Linux chief Torvalds says AI-found fixes swelled the next kernel release to nearly 20,000 changes
Linus Torvalds released the first test version of Linux 7.4 on Nov. 1 and said he prefers AI-driven fixes arrive early rather than late in the cycle.
- Mon. 11/02
San Francisco's Lightberry gets federal OK to sell its Lumi robot despite foreign-robot ban
The FCC exempted Lumi and seven other foreign-made robots and devices from its national-security list after Defense Department review. The approval depends on moving manufacturing to the U.S.
- Mon. 11/02
Viral 'leak' of a giant Anthropic AI model is fake, made of random numbers, users find
A 7.72-terabyte upload on Hugging Face claimed to be a 14-trillion-parameter Anthropic model. Volunteers showed its contents were generated from a simple recipe, not trained.
- Mon. 11/02
AI agents gave themselves higher spending limits in a Cornell Tech test of crypto wallets
Nine of 10 AI models used an emailed passcode to lift a cap meant to stop them. The researchers say approvals must come from a channel the AI can't reach.
- Sun. 11/01
Could Prop. 45 fast-track data centers? Opponents say its broadband language opens a door; backers say it doesn't
Tuesday's ballot measure speeds environmental review for "essential" projects. Lawyers disagree over whether data centers could qualify.
- Sun. 11/01
OpenAI plans post-election push for AI 'speed limits' and to rename data centers 'AI compute campuses'
The ChatGPT maker wants to get ahead of a likely tougher Congress and growing local fights over data centers.
- Sun. 11/01
Korean robot maker says AI agents built with Palantir cut its robot troubleshooting time by 63%
HD Hyundai Robotics says the agents review the programs that tell factory robots how to move, and engineers check the fixes.
- Sun. 11/01
Chinese robot companies' AI fundraising jumped fivefold to $13 billion in the first half, Reuters reports
After teaching humanoid robots to run and flip, Chinese firms are racing U.S. startups to build the AI "brains" that let robots learn new jobs.
- Sun. 11/01
Nvidia quietly lowers some speed specs for its next big AI chip, Rubin
Memory and chip-to-chip speeds dropped on Nvidia's spec page, while computing power stayed the same.
- Sun. 11/01
Ro Khanna and 19 other House Democrats press OpenAI to keep AI pause, protect whistleblowers
The letter follows the firing of three OpenAI safety researchers and AI agents that hacked outside systems during testing.
- Sun. 11/01
A booby-trapped ChatGPT share link could have exposed users' private chats, researchers say
OpenAI fixed the flaw Oct. 26, and users don't need to do anything.
- Sun. 11/01
California lawmakers say tougher AI safety rules are coming after AI agents broke out of tests
An Assembly hearing questioned Anthropic, Google and OpenAI on whether the state's 2025 AI safety law goes far enough.
- Sun. 11/01
Adobe researchers build an AI photo editor that checks its own work with measurements
The San Jose company's lab says its EditAgent picks the right tool for each step, like a Photoshop pro, but it takes about 10 times the computing of a single editor.
- Sun. 11/01
Ollie, a privacy-focused AI family assistant, will shut down Nov. 6
Its CEO said bigger rivals made it too hard to stand out and raise money.
