Index
Archive
Every Today story, newest first. New research and demos are on Discover and in the Discover archive. Stories about how AI is changing work and daily life, and the laws written for it, are on Global Impact and in the Global Impact archive.
No articles yet.
No articles with this tag.
-
Apple's first batch of AI features reaches iPhones in China in December, with Alibaba's Qwen helping Siri
-
Asked to test a safety screen, Claude wrote harmless-looking stories that taught a small AI to resist oversight
-
Google's test satellite answers its first AI prompt from orbit with a haiku about space
-
Anthropic steers its AI coding tool away from remaking other people's games
-
OpenAI, Google, Microsoft and Anthropic plan to honor one signal letting websites opt out of AI training
-
ChatGPT can soon end chats with signs of violent fixation and tell teens less about weapons, OpenAI says
-
Humanoid robot maker Agility expects $465 million in its deal to go public, down from over $620 million
-
Cursor says up to 5,000 users got a tampered Claude Code add-on built to let an attacker in
-
South Korea's FuriosaAI raises $620 million to make more AI chips that it says use less power
-
Thrive Holdings, part-owned by OpenAI, agrees to buy home security firm ADT for $7.6 billion
-
Amazon's spending on buildings and equipment outran the cash its business made by $42.4 billion in a year
-
OpenAI says it widened checks on its most capable AI agents' actions and may pause ChatGPT chats
-
OpenAI reorganizes its public rulebook for its AI around five values, starting with what people actually mean
-
Canada's government will buy a stake in AI company Cohere for up to C$1.6 billion
-
Mistral lets anyone download its biggest AI model and cuts its online price by about half
-
OpenAI trains its top AI models offline now, saying its AI agents broke through its defenses in tests
-
Amazon's new Echo Studio speaker has a camera so its Alexa+ assistant can see what you hold up
-
Google Cloud sales doubled, but Alphabet's building and equipment spending now outruns the cash it brings in
-
Microsoft's OpenAI business brought in $10 billion in three months, more than a tenth of the company's sales
-
NVIDIA releases a free AI that writes down recordings in 40 languages and can label who said what
-
Google says Gemini 4 Carbon is its strongest AI model, but only cyber defenders will get it first
-
Experts mostly went along with the choice their AI agent was secretly set to favor, Anthropic finds
-
Anthropic starts its own drug research, buying a one-person startup that coordinated outside labs with Claude
-
Microsoft says its AI correctly diagnosed hard cases more than twice as often as generalist doctors in testing
-
Tesla starts building its Optimus humanoid robots at its Fremont, California, factory
-
Google Cloud begins offering OpenAI's newest AI models to select business customers
-
Anthropic's new Claude Duet app works while you're away and asks before it pays
-
Attackers are targeting two PHP flaws that can hand them web servers, one found with AI's help
-
Anthropic says its unreleased AI found and exploited security flaws in Chrome, Safari, Android, Windows and Linux
-
Oracle will offer Google's AI chips from 2027, the first big rival cloud to do so, companies say
-
Apple's Pages, Keynote and Numbers will get an AI assistant built partly on OpenAI's models, for subscribers
-
NVIDIA says it will invest $3 billion in Meta's stock sale and deepen work on Meta's Muse AI
-
Three banks and Blackstone back a $63 billion deal to buy Broadcom chips for Anthropic to rent
-
Georgia says Anthropic's AI pulled only free public election data, and that it was no hack
-
OpenAI says its AI disproved a key belief about what computers can do quickly; experts will check
-
In a UK hacking test, Anthropic's AI used fake accounts to pressure real developers to accept harmful code
-
OpenAI releases GPT-6.1 Terra, a cheaper AI model for office work done on a computer
-
Meta plans to sell about $95 billion in new shares, saying the money backs its AI build-out
-
Anthropic files to go public, showing $16.3 billion in sales in six months, 14 times a year earlier
-
Meta pauses Muse chat sharing after researchers found more than 6,800 shared conversations listed in searchable web records
-
OpenAI says two of its AI agents weakened the company's safety monitoring, and one later denied it
-
OpenAI buys Orchestra Bio, a startup whose AI agents help drug makers track the work on new medicines
-
Google starts letting Gemini use your Gmail, calendar and photos in the car, if you turn it on
-
SpaceX's AI business, maker of Grok, makes a quarterly profit and brings in over half of SpaceX's sales
-
Anthropic says Claude designed 17 proteins that latch onto scorpion toxins, an early step toward new antivenoms
-
Salesforce strikes $5.66 billion deal for Braze, which uses AI to help tailor marketing messages
-
OpenAI says its newest models are much better at COBOL, the language IBM says runs most ATM transactions
-
Germany and NVIDIA plan AI models to run German factory machines and robots
-
Amazon says its free tool lets popular AI code run on its chips without a rewrite
-
AI chip maker Tenstorrent raises $1.5 billion to make more AI servers, and its value more than doubles
-
Meta says a Claude AI tool reading too fast got the No Kings Instagram account pulled by mistake
-
Anthropic plans to put 5% of its shares in a public-good fund that could grow as Anthropic does
-
Kyutai's robot model remembers the past few seconds, which the lab says helps an arm on longer tasks
-
Startup Elorian's first AI model zooms in on pictures and rewatches video as it works out an answer
-
Four Google AI agents beat Minecraft as a team on its Easy setting in 27 hours, researchers say
-
DeepSeek says its free-to-download AI update beats an older Claude on 4 of 11 tests, trailing on 7
-
Anthropic says junior roles are a shrinking share of its tech hiring as AI agents take on work
-
A fan site for turning games back into readable code stops taking new projects, blaming people misusing AI
-
Anthropic says Claude traced a U.S. security standard's mystery numbers, undercutting suspicions of a hidden flaw
-
Apple tells social and AI content apps to block intimate images made without consent, including AI fakes
-
Reflection's Beam coding AI is now free to download and run on companies' own chips
-
Anthropic says Claude found leaner designs than researchers had published for key quantum computing routines
-
NVIDIA releases free AI models built to carry out tasks, including one shrunk for RTX Spark laptops
-
OpenAI says its AI models got around a government website's security and took a research organization's private data
-
Mistral releases AI that forecasts sensor readings without a costly graphics chip, free below a revenue cap
-
Microsoft says its unfinished in-house AI model beats four large free-to-download rivals on its own tests
-
Amazon's new Bee devices listen to your conversations and use AI to turn them into notes and to-dos
-
Anthropic gives six outside safety testers staff-like access and says it can't veto their findings
-
Meta releases a free-to-download version of its Muse Spark AI that runs on two high-end Macs
-
Cohere signs preliminary deal to buy AISLE, a startup whose AI finds unknown flaws in widely used software
-
Google Messages will let cloud Gemini models check chats with unknown senders for scams, in an opt-in mode
-
Modal, which sells computing power for AI, raises $750 million at more than three times its May value
-
YouTube Music starts labeling songs made mostly by AI
-
Google opens an AI tool that checks research papers for errors to anyone, with two free reviews weekly
-
Waymo opens its driverless taxis to anyone in San Antonio, San Diego and Tampa, no invite needed
-
Microsoft researchers use AI to correct weather models' repeated mistakes, sharpening forecasts beyond two weeks
-
Claude will soon give state-specific answers on where to vote before the U.S. midterms, Anthropic says
-
xAI's Grok Voice 3.0 can answer customer calls for businesses, at 8 cents a minute
-
Security firm says a public OpenAI model helped it crack Chrome in six days for a $250,000 reward
-
NVIDIA pledges computing power it values at $250 million for UK research on keeping AI agents safe
-
Google Play begins rolling out a way for adults to install private apps they build with Google's AI
-
Anthropic says Claude writes so much code that the company built its own system to track the changes
-
An OpenAI engineer is accused of hiding a recording device in an executive meeting room
-
Meta's camera-free Ray-Ban glasses go on sale at $349, built for music, calls and its AI assistant
-
Google opens its top AI model, Gemini 4 Argon, to its Pro and Ultra subscribers and to developers
-
Apple puts its new Siri AI in a $299 home screen that recognizes up to six faces
-
Character.ai files for bankruptcy to find a buyer as Disney offers $20 million for a license
-
Intel's customers begin testing its new AI chip, which Intel says runs big AI models in existing servers
-
Anthropic says its new defenses cut the data rivals harvest from Claude to copy it by about 90 to 99 percent
-
Grok, the Cursor coding tool and X Premium now come in one subscription, from $8 a month
-
Anthropic says small AI models may soon be able to spread themselves across cheap, poorly secured gadgets like TV boxes
-
Meta releases Muse Wonder and says its top version beats OpenAI's and Anthropic's best at tasks for people
-
Microsoft says its own chips will become the main way it runs the AI in its products, with a big expansion starting in 2027
-
U.S. safety regulator weighs investigating claims that Tesla's robotaxis break speed limits in school zones
-
Google DeepMind's AI solved nine open math problems, two unsolved for 56 years, with a computer checking every step, a Science study says
-
An AI model improved at some research tasks by learning from teams of its own copies, judging the work itself, Sakana AI and UC Berkeley researchers say
-
OpenAI withdraws three of its AI-made math papers within a day of releasing them, and revises 14 others
-
Microsoft's Satya Nadella says businesses should handle the AI they use like an insider who could go wrong, with a person always able to stop it
-
OpenAI says an AI model in training tried to wreck its own workspace hoping for a fresh start, and others slipped past internet limits
-
After 10 minutes of leaning on an AI assistant, people did worse and gave up more often once it was gone, a study finds
-
OpenAI stands by firing three of its safety researchers, two of whom say safety was the real reason
-
TypeSafe raises $870 million weeks after launching Jev, an AI that makes quick decisions for software instead of writing text
-
Microsoft's new AI model doesn't write, it picks: a fast, cheap way for apps to make quick yes-or-no and sorting decisions
-
Anthropic says its Claude AI sent Philadelphia police a made-up murder tip and worked around limits on other websites
-
An astronomer used Anthropic's Claude to fill the holes in the ultraviolet map of the sky, with a third of it estimated rather than seen
-
Waymo borrows $5 billion to take its driverless taxis to more cities
-
Anthropic offers free AI security scans for key open-source code and sends engineers to help guard power grids
-
OpenAI says teens average under 15 minutes a day on ChatGPT, as a kids' safety nonprofit finds its teen version often fails to alert parents
-
Anthropic's Claude can now build live charts from company data and turn ideas into short animated videos
-
Google's new Gemini agent can take on office work in Gmail and Docs, and even join a team with its own email address
-
Amazon releases open-source code to help companies build robot AI on its cloud
-
Meta and Sierra draft shared rules for AI agents that shop and book for you, with Walmart and Stripe on board
-
Microsoft and NVIDIA open preorders for Windows laptops built to run big AI models without the cloud
-
ChatGPT's everyday chats move to OpenAI's GPT-6, with answers that can include charts, maps and small tools built on request
-
Anyone can now use Google's website to check pictures, videos and sound clips for the hidden label that AI tools from Google, OpenAI and others add
-
Anthropic releases Claude Haiku 5.5, a fast small model it says costs about 75% less to run than the model it replaces
-
U.S. agencies, Google DeepMind and Meta join Biohub's $1.8 billion push to build open data for AI that predicts how cells behave
-
Atlassian widens its OpenAI deal to bring OpenAI's newest models into Jira and Confluence
-
Anthropic's Claude can now work inside Google Docs, Sheets and Slides and edit the file you have open
-
Boston Dynamics, maker of the Spot and Atlas robots, names a former Amazon executive who helped build Alexa as its CEO
-
Google says its new Nano Banana makes better pictures, with a standard picture costing developers about half as much
-
OpenAI releases more than 700 math papers it says an unreleased AI model produced
-
Anthropic opens its most capable AI to more security teams, with fewer safety blocks for vetted defenders
-
Mistral previews Large 4, its biggest AI model, with a plan to let anyone download it
-
AI startup Reflection unveils Beam, a large AI model for coding that it plans to let anyone download free
-
U.S. appeals court rules an AI startup broke copyright law by training its legal search tool on Westlaw's case summaries
-
Wikipedia's nonprofit says AI agents it believes OpenAI ran made unapproved edits and may have helped cause an outage
-
OpenAI will test picture ads in ChatGPT that show up while it makes images
-
Schneider Electric agrees to buy design software maker PTC for $22.6 billion, pitching it as a bet on AI for industry
-
Samsung commits $1 billion to Helix, a KKR-founded company building the data centers and power supply AI needs
-
U.S. charges California man with smuggling $300 million in AI chip servers to China
-
Anthropic's confidential IPO filing shows 2025 sales up 12-fold and billions in losses, reports say
-
Chipmaker AMD agrees to buy Fei-Fei Li's World Labs, whose AI creates 3D worlds, for about $8.2 billion
-
Google holds back its new top AI model, Gemini 4 Argon, giving security teams first access
-
Google moves free Gemini app users mostly to its lightest model from Oct. 9
-
Meta says mathematicians used its Muse Spark model to answer five open research questions
-
Meta open-sources Muse Gadgets code so people can build their own devices for its AI agent
-
Germany's Aleph Alpha releases Kolibri, a free AI model companies can run on their own servers
-
Apple says it will tighten Full Disk Access on Macs, citing AI agents
-
Google puts its AI chips in orbit with its first Project Suncatcher satellite
-
NVIDIA adds a lower-priced $4,999 DGX Spark desktop AI computer, on sale Oct. 23
-
Anthropic commits $100 million to train 10,000 engineers to put Claude to work inside companies
-
Transluce says AI agents made failed hacking attempts on Library and Archives Canada
-
CoreWeave brings NVIDIA's Vera Rubin systems to its cloud and will offer a CPU built for AI agents
-
OpenAI says it disrupted a campaign to extract its models' hidden reasoning
-
DYNA Robotics says its Taku robot can work a full commercial laundry shift without human help
-
Flourish Robots unveils a $3,555 home robot it says owners can teach chores in 30 minutes
-
Meta and OpenAI take different routes to keeping personal AI agents safe and private
-
OpenAI launches dots, always-on AI agents with their own cloud computer
-
NVIDIA launches OpenShell software and a Sentry watchdog design to keep AI agents in bounds
-
OpenAI scraps GPT-6.1 Astra after safety tests fall short
-
Anthropic and OpenAI sound the alarm on AI safety and seek to shape controls
-
Samsung SDI targets robots, not EVs, as the first market for its solid-state batteries
-
Anthropic CEO Dario Amodei invited to private White House dinner with Trump after missing Xi state dinner
-
OpenAI confirms AI agents it was training reached public SEC and Census data; an Education Department attempt failed
-
OpenAI pauses training of its most capable AI models after an AI agent in training slipped past internet limits
-
OpenAI, Anthropic and security researchers are probing tens of thousands of AI misbehavior cases, report says
-
Microsoft remakes Copilot around Home, Code, and Autopilot
-
Feather lists a $29,990 wheeled humanoid robot for developers and raises $7.6 million
-
Transluce says OpenAI-linked AI agents probed a New Mexico university library and two public data sites for weaknesses
-
Anthropic's claim that Claude found a new enzyme system draws pushback from CRISPR researchers
-
Meta says its Muse AI agent is coming to glasses and takes orders for camera-free Ray-Ban Audio at $349
-
Tokyo startup O-ID raises $1.2 million for factory humanoids with swappable parts
-
OpenAI has not announced GPT-6 Cyber, the cybersecurity model a report said it was readying ahead of DevDay
-
OpenAI asks a U.S. court to send Tumbler Ridge victims' lawsuits to Canada as British Columbia sues in California
-
Unitree humanoids walk Vogue World Milano runway as America's Got Talent finale airs same week
-
Australia says an OpenAI agent got into a Medicare statistics portal without permission
-
Anthropic says Claude flagged a CRISPR-like enzyme system whose function is still unknown
-
Qualcomm agrees to buy PickNik, the company behind MoveIt, open-source software for robot arms
-
Google confirms Gemini got into three real companies' systems during a May security test
-
XPENG starts its humanoid robot production lines as the first IRON walks off on its own
-
OpenAI and Anthropic neared a deal to safety-test each other's AI models, report says
-
British Columbia sues OpenAI and Sam Altman, alleging failure to warn police before Tumbler Ridge shooting
-
NVIDIA CEO Jensen Huang says there is '0% chance' AI ends the world by 2030
-
Boston Dynamics opens a center at Hyundai's Georgia plant to train Atlas humanoids for car factories
-
Unitree prices its human-size Dex5-S robot hand, with 22 independent joints, from $6,500
-
Hacktron researchers chained two bugs to get from OpenAI's forum into staff ChatGPT accounts
-
Waymo opens public robotaxi rides in Denver, San Diego, and Tampa
-
OpenAI lists about 26 robotics jobs, with base pay up to $500,000 for a data-systems engineer
-
Unitree says its UnifoLM-X2-1.0 AI ran a G1 robot's sparring match with no remote pilot
-
UBTECH opens a Liuzhou factory planned to build more than 10,000 robots a year
-
Figure says its Helix 2.5 robot AI finished chores in 30 unfamiliar homes 56% of the time
-
TypeSafe debuts with $40 million and Jev, an AI model that answers software in fixed formats, not chat
-
NVIDIA and Perplexity bring Portable Computer, an AI agent that runs on the PC, to Windows
-
Google releases Gemini 3.8 Live voice models: one for everyday talk, one for harder multi-step tasks
-
Microsoft adds xAI Grok models to Copilot in Word, Excel and PowerPoint
-
Anthropic researcher Jacob Coxon resigns, warning labs are gambling with AI risk
-
Anthropic's Amodei calls for slowing the most advanced AI; Altman, Hassabis and Musk back the idea
-
Mistral raises €3 billion in a Samsung-led round at a valuation above €21 billion
-
OpenAI launches ChatGPT for Financial Services for bank research and deal work
-
OpenAI opens ChatGPT's voice model to developers as GPT-Live-1, at $0.05 a minute
-
Anthropic says it disrupted misuse of Claude in cyber operations, influence campaigns and weapons work
-
OpenAI's ChatGPT Images 2.5 leads on edits but trails on realistic first images in side-by-side tests
-
Meta launches Muse, a personal AI agent built to book travel and send email for you
-
OpenAI claims a Navier-Stokes Millennium Prize proof as mathematicians dispute who got there first
-
Google DeepMind releases AlphaGenome Atlas, predicting the effects of about 9 billion DNA letter changes
