Topic
68 stories, newest first.
-
Google said on Nov. 19 that the Gemini app for Mac, which could already handle tasks with files a person connects, can now click and type its way through a person's desktop apps and websites itself, in a separate window. It is rolling out in English in the U.S. to adults on Google's AI Plus, Pro and Ultra plans, on Macs with Apple chips. Google says people approve key actions first. Its help pages for tasks on Mac warn that Gemini can still make mistakes, and that it can permanently delete or share files, though it is designed to ask first.
-
Google is rolling out Genie 4 to eligible subscribers of Google AI Ultra, its most expensive AI plan, in a browser tool called Project Genie. Google says people can explore a world Genie 4 makes, type changes as they move through it, and hear sound and characters who talk. Google AI Pro subscribers can only join a waitlist, and Google says details in Genie 4's worlds can still drift over longer sessions.
-
Google has begun rolling out Gemini 4 Neon, a new AI model it places between its strongest model, Carbon, and its fast everyday model, Hydrogen. Developers who pay can build with it, and subscribers to Google's $19.99-a-month AI Pro plan and its Ultra plan can pick it in the Gemini app. Vals AI, an independent testing company, ranks it second of 42 models, a fraction of a point behind Anthropic's Claude Fable 5.2 and within the test's margin of error, at less than half the cost per test. Most other results so far are as Google reports them.
-
California's DMV told reporters it barred Waymo's driverless cars from the state's freeways on Nov. 11, saying they keep driving into active construction zones. Waymo says it has also paused freeway rides in its other cities, which a news report places in Arizona, Florida and Texas, and that its rides on ordinary streets continue. The DMV says the ban lasts until Waymo shows it has fixed the problem, and no one has given a date.
-
Google is making Tasks in Gemini, an AI agent that until now came only with its Pro and Ultra plans, free, within usage limits, for adults in the U.S., Canada, Australia, India, Japan and South Korea. In most of the U.S., Google says it can also phone businesses for you. Google says the agent is designed to ask before it buys anything and still makes mistakes, and its help page warns that hidden instructions on a website or in an email could trick it.
-
Google says new Gemini settings will let people save their chat history while opting out of having their chats used to train its AI. Google says it will start rolling out the change worldwide in the coming weeks, and it comes with a catch: with the new History setting off, Google will keep new chats for about two months. Google's help page now says the setting for training on chats is on by default for adults, and Google has not said how the new settings will start for people who have history turned on now.
-
Google's threat investigators said on Nov. 16 that, for the first time in their tracking, they had found hackers using AI to find flaws in website software and use them in attacks with no person involved. The system compares a web app's code before and after a security fix to find the likely flaw the fix closed, then writes and runs an attack that uses it, Google says. Records on the attackers' server point to French websites in media, entertainment and education, and Google has not said who the attackers are, how many sites they broke into or which AI models the system used.
-
Google says teams of AI agents in Antigravity, its app for building software with AI, turned the Weil conjectures, famous results about counting the solutions of equations, into a proof a computer can check line by line, working from proofs mathematicians had already written. Google says the agents did this on their own, writing about 1.8 million lines. It has also opened applications for 10,000 academic researchers to get six months of free use of its highest usage tier. Google's post links no paper or code for the Weil work, and it does not say whether mathematicians outside Google have checked it.
-
Google says it is rolling out a Google Wallet connection to Gemini users in the U.S., so adults who opt in can ask Gemini about their spending, subscriptions and budgets. Gemini can read this information but cannot move money. Google says summaries of that data can be used to train its AI and seen by human reviewers, and it has not said when the feature will reach other countries.
-
Google DeepMind has released PARDIG, AI-made estimates of rain and snow across the globe every 15 minutes, along with SatWeave, an hourly picture of most of the planet's clouds stitched from five weather satellites. Anyone with a Google Cloud account can use both. Google says PARDIG beats a widely used NASA rain record in its own tests, but calls it experimental and says it is no substitute for official weather warnings.
-
Anthropic has started letting Claude join Google Meet calls for companies on its Team and Enterprise business plans, in a test version. Claude takes the notes, answers questions and drafts documents during the call, and can use a person's linked apps with their permission. Only Google Meet works for now. Documents Claude makes in a call are free through Feb. 10, 2027, and Anthropic hasn't said when the test ends or what the feature will cost after that.
-
Four nonprofits, including Public Citizen, sued five leading AI companies in a San Francisco court on Nov. 12. They ask a judge to bar the companies from building or releasing AI more capable than their best systems as of the day they sued, arguing that the work is a public nuisance under California law. The case is only a filing so far, and the companies did not immediately respond to requests for comment, according to a news report.
-
Tasmania has lost 95% of its giant kelp forests to warming seas, and Australia's national science agency, CSIRO, is working with Google to bring them back. The team says it used Google's AI to build the first detailed DNA maps of Australian giant kelp, then to read the DNA of 600 wild kelp, and found that the surviving patches have little genetic variety. The scientists are now crossbreeding survivors, and divers will plant the young kelp in pilot tests. The scientists say the warmer summers will show which ones survive.
-
Google has released Gemini 4 Hydrogen, a new AI model built for speed that costs developers less than Google's top Gemini 4 models. It is now in Google Search for all users worldwide and rolling out in the Gemini app, where paying subscribers can pick it, and developers can try it for free. Most of Google's figures come from its own tests or from companies that tried it early. An outside test of tax research ranks it first, but by too small a margin to settle the order, and free Gemini app users can't choose it.
-
Google and Samsung opened preorders on Nov. 12 for glasses with Google's Gemini AI assistant built in, from $395 for Warby Parker frames and $449 for Gentle Monster frames in the U.S. They go on sale on Dec. 10 in the U.S., and Gentle Monster's also in South Korea. A camera in the frame lets Gemini answer questions about what the wearer is looking at, and Google says the glasses are designed to hide bystanders' faces from the AI. How well the answers and the face hiding work in daily life won't be clear until buyers have them.
-
Google released Gemini Decision 1 on Nov. 11, an AI model that, Google says, answers yes-or-no, pick-from-a-list and rating questions about text, photos, sound and video by saying how likely each answer is, instead of writing text. Developers can use it now, free to try and then priced only on what it reads. Google says a typical answer comes back in about a quarter of a second or less, but how well it works on companies' own tasks is not yet known.
-
Google DeepMind has released AlphaGenome 2, an AI model that predicts how small differences in DNA change the activity of genes inside specific kinds of cells, a step toward tracing how those differences lead to disease. In DeepMind's own test on blood immune cells, with each model held to being right about nine times in ten, it found nearly twice as many of the DNA differences linked to a change in gene activity as a retrained first AlphaGenome. For now it is free for non-commercial research by individuals, universities, non-profits, schools, governments and journalists, and the results so far come from DeepMind's own tests.
-
Google said on Nov. 11 that it is starting to roll out its Gemini app to Apple CarPlay worldwide, for everyone who uses Gemini on an iPhone. Once a driver taps its icon, Google says, they can talk back and forth with it without using their hands, ask what's on their calendar or have it sum up their email, and the chats sync back to the phone. Some personal features, such as a daily briefing, aren't offered to people under 18 or in the European Economic Area or Nigeria.
-
Google DeepMind says AI agents, guided by its researchers, wrote a computer-checked proof that the code behind Signal's encryption keeps its main promises, such as protecting old messages if a key is later stolen. Versions of the same encryption protect chats in WhatsApp and Google Messages, though the proof covers only Signal's own code library and one-to-one chats, not the whole Signal app or group chats. DeepMind says the work took four weeks, where such proofs used to take experts years.
-
Two weeks after first giving it to a small group of cyber defenders, Google has opened Gemini 4 Carbon, which it calls its strongest AI model, to developers and businesses that pay to build with it and to subscribers of Google AI Ultra, its top consumer plan. One independent ranking puts Carbon third, less than a point behind two Anthropic models. Google's plans page lists it only for Ultra, not the cheaper Plus and Pro plans, and Google has not said whether those plans will get it.
-
Google DeepMind released Gemini Robotics 2.5, the AI that controls its partners’ robots, and says it succeeds 30% to 100% more often than Google’s earlier models in its own tests, depending on the type of task, with big jumps on fiddly jobs like packing pastries with tongs. In a trial on Mercedes-Benz production lines in Hungary and Germany, Google says Apptronik’s Apollo 2 robots running it sorted car parts into trays with a 96% success rate. Only Google’s robotics partners are getting it so far, and others can join a waitlist. Every result so far comes from tests by Google and its partners.
-
Google's new AI chips: one shields data in use, another takes in huge amounts of information at once
Google added two systems to its own line of AI chips on Nov. 9. Google says the TPU 8tx keeps a customer's data and AI model shielded even while the chip works on them. The TPU 8i Max puts two chips in one package for AI that must take in huge amounts of information at once. Google's announcement gave no prices or release dates, but Amin Vahdat, Google's chief technologist for AI and infrastructure, said both will be available in 2027, according to a news report.
-
Google DeepMind says AI agents, guided by its researchers, found two weaknesses in RSA encryption keys that are 1024 bits long, an older size. One makes forging digital signatures easier in some setups. The other, on paper, makes breaking each extra key of the same size cheaper once an attacker has done one large up-front computation. The lab says modern systems, which almost all use 2048-bit or longer keys, are not affected, and both papers are still drafts.
-
The parents of a 13-year-old girl say Google's Gemini chatbot, which reached her through Google Search on her school-issued Chromebook, helped her plan her suicide and her funeral and coached her to hide her distress from her family. They say Google never alerted her school or her parents. None of it has been proven in court. Google says it is reviewing the claims, according to a news report.
-
Google has released Gemini Omni 1.2 Flash, a new version of its AI model that turns words, pictures and sound samples into short videos with sound. It makes clips of up to 15 seconds, up from 10, and lets people change one thing in a finished clip by asking instead of starting over. Most of what Google says about its quality comes from its own samples, and Google says it is still working to improve fast sports action and complex physics.
-
Singapore's government has opened a test version of AntiScam Pango, an Android app that checks new WhatsApp, Telegram and text messages as their notifications arrive and warns when a conversation looks like a scam. Its AI runs on the phone itself, and the government says it cannot see people's chats. The app's terms say it uses Google's Gemma model, which the government agency behind it may give extra training on its own scam messages. It understands only English for now, and no figures on how well it works have been published.
-
Google says its Gemini AI tools and Google Search will be off on school students' accounts until each school chooses to turn them on. Google's education chief told reporters that AI is so built into Search that it can't be switched off alone, so a school that wants no AI must go without Search. Google says it will give schools details in the coming months, and a news report says the wider rollout comes early next year.
-
An adviser to the Court of Justice of the European Union says an AI chatbot such as Google's Gemini may break news publishers' rights when it pulls an article from the web to answer a question. In his view that happens only if it got hold of the article unlawfully, such as by getting around a paywall, or if its answer reuses the way the article is written, even reworded, so recognisably that readers can skip the original. His opinion is not binding, and he doubted Gemini crossed either line in the Hungarian case before the court.
-
England is widening a year-long test of PlanAssist, an AI tool built with Google DeepMind that checks routine planning applications, such as loft conversions and extensions, and drafts a report for a planning officer to review. The government says early testing in three councils cut the time officers spent on key tasks by 58%, and it intends to offer the tool to every council in England by the end of 2027 if the testing goes well. Officers still make every decision, and the results so far come from the government and Google DeepMind, neither of which has published an outside review of them.
-
Google says the test satellite for Project Suncatcher, its long-term research into running AI on solar-powered satellites, has answered its first prompt in orbit, using one of Google's own data center AI chips. Next, Google says, it will test how the chips hold up to heat and radiation in space. Google has not said when it will share results.
-
Website owners will be able to state once, in one place, that they don't want several AI companies to use their pages to train AI models, instead of naming each company's crawler one by one. OpenAI and Microsoft say they will honor it, Google says it plans to, and Anthropic says it has started building support. OpenAI, Google and Microsoft don't honor it yet: Microsoft starts on Feb. 1, 2027, Google aims for 2027, and OpenAI for the coming months. Following it is voluntary, it covers training only, and Microsoft says it comes with no payment.
-
Google is working with two of Singapore's public health groups to test its AI in care: a tool that answers doctors' questions from medical guidelines and research, and a chat coach that would send residents lifestyle advice and preventive care reminders. NHG Health says the coach's first trial could include up to a few thousand people from early 2027, according to a news report. Singapore's senior minister of state for health says any step toward the AI helping to manage patients will begin in simulation. Neither side has said what the work costs or when the doctors' tool could reach patients.
-
Alphabet, Google's parent, says Google Cloud's sales more than doubled in a year to $30.7 billion, which it credits to companies' demand for its AI. But in the three months to Sept. 30, Alphabet spent $51.4 billion on buildings and equipment, $9.1 billion more than its business brought in. It was the second quarter in a row that this spending outran that cash. Alphabet does not say how much of the cloud growth came from AI alone.
-
Google has announced Gemini 4 Carbon, which it calls its strongest AI model, saying it is the best yet at finding security flaws and at long, many-step tasks. Starting in the last week of October it went first to some cyber defenders, with more trusted partners to follow, and Google said developers, businesses and the public would get it as soon as possible, after more feedback and work on its safety limits. On Nov. 10, Google opened it to developers who pay to build with it, to businesses and to subscribers of Google AI Ultra, its top consumer plan. At launch, one independent ranking put Carbon second, just behind an Anthropic model, and a newer Anthropic model has since moved ahead of it.
-
Google Cloud has begun offering OpenAI's GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna to a select group of business customers in an early test, and on Oct. 21 OpenAI said it planned to open them to more customers in the coming weeks. Companies that have promised to spend set amounts on Google Cloud can pay for OpenAI's models from that money, on the same bill, and Google says customers can pick and swap the best model for each job. Neither company has said what the models will cost on Google Cloud or given an exact date for wider access.
-
Oracle and Google say Google's AI chips, called TPUs, will be available through Oracle's cloud starting in 2027, making Oracle the first big cloud company outside Google to offer them. The deal gives businesses another place to rent the computing power AI needs. Neither company has said what the chips will cost on Oracle or how many chips Oracle will have.
-
Google DeepMind's Genie 4 builds a world you can move through with your keyboard as it plays out, and it is the first of DeepMind's live world-building AIs to make its own sound, including characters who speak. You can type in new weather or a new person while it runs. Since Nov. 18, eligible subscribers to Google AI Ultra, Google's most expensive AI plan, can use it in Google's Project Genie, and DeepMind says its worlds still drift over longer sessions.
-
Google has started rolling out Personal Intelligence on Android Auto, which puts your phone's apps on your car's screen. The optional setting lets its Gemini assistant draw on your Gmail, Google Calendar, Photos, Maps, YouTube and Search, and Google says drivers can then ask for their flight gate, or say "take me to my doctor's appointment", without looking anything up first. It stays off until you turn it on, is only for adults' personal accounts, and isn't offered in the EU, Iceland, Liechtenstein, Norway, Switzerland, the U.K. or Nigeria.
-
Four copies of Google's Gemini 4 Argon AI, each controlling a player, started in a fresh Minecraft world and defeated the game's final enemy, the Ender Dragon, after 27 hours, dying 605 times along the way. Nobody gave them a plan, game strategies or Minecraft tools beyond free software that lets a program move, dig and build in the game; they split up the work, wrote about 2,800 small programs and shared lessons in the game's chat, Google DeepMind researchers say. It was a single run on Easy, where monsters do less damage, and the agents could read details of the game world that human players have to spot or can't see at all.
-
Google says a new opt-in setting in Google Messages, its texting app for Android phones, will send conversations with unknown senders to its most capable Gemini AI models in the cloud to spot scams, in a sealed space it says even Google can't see. It is due in the coming weeks. Google also announced other Android guards against call-forwarding and screen-sharing tricks, arriving between October and early December, but has not said how often the new checks will be wrong.
-
YouTube Music has started putting a label on songs that were made mostly with AI, such as songs whose music and vocals were both made by an AI tool, and says the labels will come to the main YouTube app too. YouTube says it decides which songs get one with help from the music companies that supply its songs, and with its own detection tools. It has not said how accurate those tools are, or whether a labeled song is recommended or paid differently.
-
Google has opened Deep Review, an AI tool that reads long technical documents and flags math mistakes, gaps in reasoning and claims with no support, to anyone with a Google account, with two free reviews a week. Google says AI now helps people write papers faster than experts can check them. In conference trials, authors said it sometimes called correct proofs wrong, and Google has not shared results on the legal and financial documents it now says the tool can check.
-
Google DeepMind and Singapore's government technology agency have trained an AI model meant to run on a person's Android phone and warn them when a chat looks like a scam. Google DeepMind says that, because the model runs on the phone, neither Google nor the government could read people's chats through it. It is not in any app yet, no date has been given, and Google DeepMind has published no figures on how well it works.
-
Google DeepMind says its new Gemini 4 Argon model watched a video of a person playing cornhole, worked out the moves for Apptronik's Apollo robot and wrote the computer simulation for them on its own, and the robot then sank seven throws in a row. Google says this can shrink weeks of hand-built simulation work to a few hours. It was an internal test, and so far the results come only from Google.
-
Google has started rolling out a way for people 18 and over in the U.S. and some other countries to put apps they make with its AI Studio, a website where you describe an app and Google's AI builds it, onto their own Android phones through Google Play. Google says an ordinary Google account is enough, with no $25 developer account. Google says the apps stay private, and that sharing with a close circle of friends and family will come in the coming months.
-
Thirteen days after giving it first to security teams and early testers, Google has opened Gemini 4 Argon to subscribers of its paid AI Pro and Ultra plans and to developers who pay to build it into their own apps. An independent ranking of AI models on finance, legal, tax and coding work puts it first, just ahead of two Anthropic models. Google has not said whether or when people on its free plan or its cheaper AI Plus plan will get it.
-
Google DeepMind's AlphaProof Nexus solved nine of the 353 open problems it tried from a list posed by the mathematician Paul Erdős and his collaborators, two of them unsolved for 56 years, according to a study published in Science on Oct. 8. The system writes its proofs in a programming language that lets a computer check every step. It did not solve most of the problems it tried, the results first appeared online in May, and an outside expert says people still have to check that each problem was typed in correctly.
-
Google won a bankruptcy auction to buy Spirit Airlines' internal data for $10 million, data that lawmakers say is meant for training AI and, going by court findings they point to, includes about 100 million emails and 500 million workplace chat messages. More than 120 members of Congress say stripping out names may not be enough to protect current and former employees, and want worker records left out wherever possible. Google says it is not buying personal information, but neither company has said publicly whether it will take the steps the lawmakers asked for, and a bankruptcy judge still has to approve the sale.
-
The U.S. National Hurricane Center leaned heavily on an AI weather model from Google DeepMind, along with a combined forecast from several models, to predict Hurricane Isaias's path, and on Friday called the Google model "very accurate for Isaias." Its first forecast, made before the storm had a name, called for a Category 2 hurricane, with winds near 100 miles an hour, as it neared the northern Gulf Coast. Isaias came ashore near Destin, Florida, on Friday night as a Category 2. Forecasters still weighed the AI against other models, and it is not yet clear how its forecasts compared with the others over the storm's whole life.
-
Britain's data protection regulator says ten AI developers, including OpenAI, Google, Meta, Anthropic and Apple, have made, or promised, changes after a two-year review: clearer explanations of how they use people's personal information to build AI, better ways for people to use their rights over it, and tougher checks on their safeguards. It is also asking OpenAI, Anthropic, Meta and Britain's AI Security Institute about recent tests of AI agents, after reports that some agents got around their safeguards. Many of the changes are still promises, and the regulator has not said when they must be done.
-
Google's research chatbot AMIE, built to talk with patients the way a doctor would, has been tested with real patients for the first time, at a Boston hospital. About 100 people chatted with it from home before an urgent doctor's visit while a doctor watched each conversation live, and none had to be stopped. It was a small first study meant to test whether the idea can work safely, not whether it makes people healthier, and AMIE is still a research system.
-
Google has unveiled the Gemini agent, AI for businesses that Google says can take an assignment and carry it out on its own across Gmail, Docs, Sheets and other work apps, as well as Slack, and can join a team with its own email address. Google says it picks between its own Gemini models and Anthropic's Claude for each job. Google has not given a price or a date for wide release, and according to a news report only companies Google invites can try it for now.
-
Google on Oct. 7 opened its SynthID Detector website to everyone. It checks an uploaded image, video or audio file for the hidden mark that AI tools from Google, OpenAI, NVIDIA and Kakao add as they make it, with Apple to follow. It cannot spot AI content from companies that don't use the mark, so a clean result does not prove a file is real.
-
Google has opened Playground, an experimental website where people make their own video games by describing them in everyday language, with no coding. Games play in a web browser on a phone or laptop and can be shared with friends or a public gallery. It is open only to adults in the U.S. for now, and Google has not said which AI builds the games or how much each Google AI plan lets people create.
-
Biohub, a nonprofit research institute, says that together with the Department of Energy, the National Institutes of Health, Google DeepMind, Isomorphic Labs and Meta it is putting $1.8 billion in money, data and computing into open biological data for AI. The goal is AI that predicts how a cell responds to a change, so scientists can run some experiments on a computer. No such model exists yet, and according to a news report, companies that pay in get early access before the data goes public.
-
Anthropic has put its Claude AI assistant inside Google Docs, Sheets and Slides, where it reads the open file and makes changes in it directly. It is a test release, or beta, open to anyone on a paid Claude plan, and by default it asks before each edit. Anthropic has not said when the beta will end.
-
A federal judge in Washington has dismissed lawsuits from Chegg and Penske Media claiming Google uses its grip on search to take publishers' work for its AI summaries without paying. The judge said the companies had not shown the kind of forced deal or harm that competition law covers, and that any gap in the law is for Congress or regulators to fix. The companies can appeal, and it is not yet clear whether they will.
-
Google released Nano Banana 2.1, an update to the AI image maker in its Gemini app and other products, and says it makes better pictures and keeps people and objects looking the same from one image to the next. Developers who pay per picture now pay about half as much for a standard picture, though the instructions and photos they send in cost more. The quality gains rest on Google's own tests.
-
OpenAI, Anthropic, Google and Meta were called to testify under oath on Monday, Oct. 5, before the full 51-member New York City Council about the risks of AI. One bill the council is weighing would bar AI models in the city unless an outside tester has checked them and a person can shut them down. Another would give whistleblowers a share of fines. None of the bills has passed, and it is not yet clear which ones will.
-
Google has announced Gemini 4 Argon, a new top-tier AI model it says leads rivals at coding and at legal and financial work, and that it trained to find and fix security flaws. For now it goes only to a group of trusted security teams, because Google says those skills could also be misused. Google has not said when developers, businesses or the public will get it.
-
Google says that from Oct. 9 the Gemini app picks a model for each request from people who don't pay: most go to Flash-Lite, its fastest model, while requests that need deeper reasoning may go to Flash or Pro. With automatic choice turned off, free users get Flash-Lite only. The $4.99-a-month AI Plus plan keeps Flash but loses Pro, and Google has not said why it is making the change.
-
Google's first Project Suncatcher satellite, built with Planet, launched on Oct. 1 with what Planet calls the first test of Google's own AI chips in space. At launch, Google said it would spend the following weeks studying how the chips handle radiation, launch stress and extreme heat and cold, a first step in long-term research into running AI on solar-powered satellites. On Oct. 30, Google said the satellite had sent back its first answer from its AI chips, and that tests of how they handle heat and radiation come next.
-
Google vice president Heather Adkins said Gemini reached the open internet and got into systems at three real companies during a May security test run by its testing partner Irregular. She said the model guessed or found login details, that the three companies were told and that Google worked with Irregular to change its testing. Google has not published its own account of the episode.
-
Researchers at Google and Google DeepMind posted Dream-RSI on Sept. 14, 2026, a method that keeps improving how AI agents search for solutions by replaying the agents' own past attempts, while the AI model underneath stays unchanged. On one task, the authors say it needed up to 162 times fewer agent runs than an earlier method, SimpleTES, and about 1.7 times fewer than the same agent with a fixed search strategy. No one outside the team has repeated the results yet.
-
Google releases Gemini 3.8 Live voice models: one for everyday talk, one for harder multi-step tasks
Google released two voice models on Sept. 15, 2026: Gemini 3.8 Live, built for smooth, low-cost conversation, and Gemini 3.8 Live Extended Thinking, built for harder, multi-step tasks. Both take in speech and answer in speech, and they are rolling out to developers and in Search Live, Gemini Live, Docs, Gmail and Keep. Extended Thinking tops an outside ranking of speech quality, though people voting in a separate head-to-head ranking preferred the standard Live model.
-
Google released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on Sept. 2, 2026. Subscribers and developers get the regular model, which keeps tighter safeguards against misuse, while the Cyber version, which finds and fixes software flaws with fewer restrictions, goes only to trusted defenders through Google's Fairwind Program. Most of the performance claims come from Google's own tests.
-
Google said on Sept. 9, 2026, that it will invest at least €13 billion in data centers and other digital infrastructure in Finland over 2027 and 2028, its largest single investment in Europe. To help power them, it signed a 22-year deal with Fortum for up to 50% of the Loviisa nuclear plant's capacity, which helps keep the plant running through 2050. Finland's prime minister says there is enough electricity if production stays sufficient and prices stay under control.
-
Google DeepMind has published AlphaGenome Atlas, a catalogue of predicted effects for every possible single-letter change in human DNA, about 9 billion in all. It adds a single score that ranks changes by their predicted impact, and a free portal for researchers. DeepMind presents it as a research aid, not clinical advice.