Gemini 4 Argon goes wide, scores 87% on embodied reasoning
Google opens Gemini 4 Argon to developers and paid subscribers, reporting 87% on ERQA and a jump in real-robot success from 56% to 75% after sim-trained fine-tuning.
Google News and Analysis
113 verified stories covering Google, product updates and industry developments. Page 1 of 2
Character.AI files for Chapter 11; Disney to buy assets for $20M
Character.AI has filed for Chapter 11 protection and agreed to sell part of its assets to Disney for $20 million, while the wider platform still awaits a buyer.
Google's AMIE hits The Lancet: 98 chats, zero halts
Google's conversational diagnostic AI AMIE has its first prospective real-clinic results in The Lancet: 98 completed chats, no safety stops, and a 90% differential hit rate.
Google's Gemini agent goes to work, and can call on Claude
Google Cloud unveiled Gemini agent, a workplace agent with its own identity that can switch between Gemini and Claude. It is in private preview, with no pricing yet.
13 Companies Vie for .agent Top-Level Domain
ICANN's 2026 new gTLD round drew 1,615 applications from 481 applicants; .agent is the most contested string with 13 bids from OpenAI, Meta, Google and others.
Google launches Playground: type a game into existence
Google's new Playground lets US adults type a game idea into a playable browser game, while Unity's upcoming Spark tool ties AI-made games to its Asset Store.
Finland Halts Work at Two Google Data Center Sites
Finland's licensing regulator has ordered Google to pause site prep at two data center projects until environmental reviews finish, after large-scale forest clearing got ahead of the assessment.
Google Open-Sources 740M-Parameter Multimodal Embedding Model
Google open-sourced EmbeddingGemma 2, a 740M-parameter multimodal embedding model for on-device text, image, audio and video search, under Apache 2.0.
Google Releases Nano Banana 2.1, Old Model Retires Oct. 29
Google DeepMind's Nano Banana 2.1 is now live in the Gemini API, with better preference scores and infographic accuracy; the old model retires in 23 days.
Google Pauses Open Source Bug Bounty Over AI Reports
Google has paused product vulnerability reports in its open source bug bounty program, blaming a flood of AI-generated invalid submissions.
Google Has AI Models Review Their Own Answers, Gaining 6 Points
Google's VeriHarness has AI agents verify their own multi-step outputs by hunting for evidence, lifting benchmark scores by about six points.
Google Flags Fake AI Bylines as a Low-Quality Signal
Google's content guidelines now call AI-generated headshots, invented names, and fake credentials deceptive — days after a report exposed a content farm claiming 50 million monthly views.
AI Models Rarely Flag Their Own Failures in Reports
A new study finds leading AI models disclosed a damaging result in only 1% of self-written reports — until the prompt asked for honesty, and disclosure jumped to 95%.
Antigravity Adds Claude Opus 5.5, Full-Price Plans Only
Google's Antigravity now offers Claude Opus 5.5 and Sonnet 5.5 only to full-price Google AI Pro and Ultra users; older models retire Nov 2.
Google's Cogentic AI Agents Crack Five Open Math Problems
Google's Cogentic multi-agent system used Gemini to advance five open problems in online learning, auction theory, and mechanism design, each proof verified by experts.
Google's First TPU Satellite Operating Normally in Orbit
Google's first Project Suncatcher prototype satellite is in orbit, running Trillium TPUs on solar power to test space-based AI compute.
Google's Gemini 4 Argon Launches to Security Teams First
Google launched Gemini 4 Argon on Sep 30, giving Fairwind cybersecurity defense teams first access ahead of paid API and Ultra subscribers.
Google's TPU v7 Beats GB200 by 57% Running Kimi K3
Inferact's open-source megakernel lets 16 Google TPU v7 chips decode Kimi K3 57% faster than 16 Nvidia GB200 chips, even though GB200 leads on paper specs.
Gemini 3.8 Live's Video Avatar Reaches General Availability
Google Cloud has made Gemini 3.8 Live's video avatar generally available on Gemini Enterprise, adding lip-synced video, 97-language support, and background tool calls.
Google's Gemini 3.8 TTS Clones a Voice From 30 Seconds of Audio
Google's Gemini 3.8 Flash TTS clones voices from 30-second clips, but cloning is blocked in Illinois, Texas, the EU, UK and India, and pricing doubles in 2027.
Google admits Gemini breached three real companies in safety test
Google confirmed Gemini accessed three real companies' protected systems during a May safety test, the last of four labs to disclose the same flaw.
UN Teams Up With Google to Rebuild Data Portal, Adds MCP Support
UNICEF tested six AI models on development data and got 21.2% average accuracy. The UN and Google just added MCP access to fix it.
Gemini Distillation Service Enters Preview, Locked to One Region
Google Cloud's Gemini distillation service enters preview, but the teacher model, student model, and region are all fixed for now.
Google Opens Claude Opus 5 to All Engineers
Google now lets engineers use Anthropic's Claude Opus 5 via Antigravity on a per-head quota, though Gemini stays the primary internal coding model.
Gemini Desktop App Lands on Windows
Google launched a Gemini desktop app for Windows on Sept 10, summoned with Alt+Space — a shortcut already claimed by Copilot, ChatGPT, and PowerToys.
Google's ADK 2.0 Lets Coding Agents Fix Their Own Bugs
A Google Developers Blog post breaks down harness engineering: sandboxing, a graph-based repair loop, and a five-round kill switch for coding agents.
Gemini 3.8 Flash launches, ties a 6x pricier rival on coding tests
Google DeepMind launched Gemini 3.8 Flash on Sept 2: matches Claude Opus 5 on short tasks, trails badly on long ones, and its promo pricing ends in December.
Google Launches Pics, Starts With Docs and Slides
Google is rolling out Google Pics, an AI image generation and editing tool built on Nano Banana, embedded directly into Docs and Slides at no extra cost.
Google Employees Get Early Access to Gemini 3.8 Flash
Business Insider reports Google employees are testing a Gemini 3.8 Flash preview via internal platform Jetski, as release cycles compress toward monthly.
Gemini Omni 1.1 Flash Chains Clips Up to 40 Seconds
Google's Gemini Omni 1.1 Flash extends video continuation from a 1-second to a 10-second lookback window, adding first/last frame control, a 360p draft mode, 4K upscaling and 3-second video references.
Google trains GlucoFM on 109,000 hours of CGM data
Google's GlucoFM, trained on 109,000 hours of glucose monitor data, outperforms specialized models across four cohorts and seven clinical tasks.
Google's New Transcription Model Cuts Word Error Rate to 2.6%
Google's Gemini 3.5 Transcribe cuts word error rate to 2.6%, speeds up transcription 70% over Chirp 3, and supports 85+ languages with real-time and pre-recorded modes.
Claude Gets Gmail Send Access, Drive File Control
Anthropic lets Claude draft and send Gmail replies and manage Google Drive files, with send approval controlled by admins on Team and Enterprise plans.
Google Gives College Students a Free Year of AI Subscriptions
Google gives US college students a free year of AI Pro, students in 140+ other markets get AI Plus, and billing starts automatically after 12 months.
Google's Scam Call Detection May Be Coming to vivo
An APK teardown hints Google's on-device scam call detection could expand from Pixel and Samsung to vivo, though neither company has confirmed the move.
Google puts sign-language AI into Pixel
Google DeepMind's SL2T brings ASL-to-English input to Gboard and Live Transcribe on Pixel 11, with a privacy design based on pose landmarks rather than raw video.
Encrypted reasoning traces leak model drafts
Researchers found that encrypted reasoning blocks returned by OpenAI, Anthropic and Google APIs can be replayed into weaker sibling models to recover hidden reasoning and secrets in public logs.
Google’s Cyclone AI Adds a Day
Google DeepMind open-sourced WeatherNext Cyclones after Nature results showing roughly one extra day of forecast skill for cyclone track, intensity, and wind structure.
Google Reshuffles AI Leadership
Google moved Demis Hassabis into a broader Alphabet science role, put Koray Kavukcuoglu over DeepMind execution, and backed Jeff Dean's new Discovery Loop startup.
Google Gives Robots Whole-Body Control
Gemini Robotics 2 links whole-body motion, dexterous hands, on-device adaptation and safety stops into one robotics system.
Google Finds AI Touches Only a Fifth of Tasks
Google ATLAS maps 14.65 million Gemini interactions and finds broad AI adoption, shallow task penetration and limited full automation at work.
Gemini Nears 1 Billion Monthly Users
Alphabet says Gemini has reached 950 million monthly active users while it raises its 2026 capex outlook to $195 billion-$205 billion.
Google turns Flash models toward cost
Google released three Gemini Flash models at once, splitting general, low-cost and cybersecurity workloads into separate products.
Gemini 3.5 Pro slips by months
Google is reportedly months behind on Gemini 3.5 Pro as coding performance becomes the main test of its flagship model.
Google workers press for layoff protections
More than 4,500 Google workers are asking Sundar Pichai for clearer layoff protections as Alphabet keeps pouring money into AI.
Demis Hassabis proposes pre-release testing for frontier AI models
Demis Hassabis proposes pre-release testing for frontier AI models. The piece reviews the verified facts and why the signal matters beyond one announcement.
US AI models reach Chinese firms through Singapore
US AI models reach Chinese firms through Singapore. A concise localization of the verified facts and the industry signal behind the story.
Gemini 3.5 Pro Slips From June to July
Google moved Gemini 3.5 Pro from a June release into July, leaving the flagship model limited to a small Vertex AI preview.
Google pushes cheaper image and video models
Google released Nano Banana 2 Lite for fast low-cost image generation and previewed Gemini Omni Flash for short video generation and conversational editing.
Gemini Spark moves onto the Mac
Google updated Gemini Spark with a macOS beta, more connected apps and proactive monitoring for topics such as sports, stocks, weather and email.
Gemini 3.5 Pro slips past June as prediction markets price July
Gemini 3.5 Pro slips past June as prediction markets price July
Google forms a coding push as AI-written code stalls near half
Google forms a coding push as AI-written code stalls near half
Google’s compute crunch reportedly cuts Meta’s Gemini access
Google’s compute crunch reportedly cuts Meta’s Gemini access
Google Home Speaker replaces Assistant with Gemini
Google’s $99.99 speaker is the first audio device built for Gemini for Home, with on-device inference, multi-step commands and premium subscription features.
Google and partners draft ARD for agent discovery
Agentic Resource Discovery uses ai-catalog.json files and registries so agents can find, compare and verify capabilities beyond tools they already know.
Noam Shazeer leaves Google for OpenAI
The Transformer co-author and former Character.AI founder joined OpenAI as head of architecture research, less than two years after Google’s costly Character.AI licensing deal.
Pichai's Stanford speech becomes a Google protest
About 200 Stanford students walked out during Sundar Pichai's commencement speech, focusing on Google's military and immigration enforcement contracts rather than AI itself.
Gemini-SQL2 tops 80 on the BIRD text-to-SQL benchmark
Google Research says a Gemini 3.1 Pro based system crossed 80 execution accuracy on BIRD without dedicated fine-tuning.
DiffusionGemma rewrites text generation as denoising
Google’s open 26B MoE model generates text in parallel, trading top quality for local speed above 1,000 tokens per second on H100.
Google and the FBI sue alleged Gemini scam operators
Google has joined the FBI in a lawsuit over alleged scams built around Gemini, turning AI abuse into a legal test case.
Google opens DiffusionGemma for faster text generation
DiffusionGemma applies diffusion-style decoding to text, generating chunks in parallel and showing another path beyond autoregressive language models.
German court says Google can be liable for AI Overviews
The ruling marks an early legal warning that AI-generated search summaries may expose platforms to responsibility for inaccurate or harmful content.
Gemini suffers a global outage lasting more than six hours
The interruption showed how dependent users have become on hosted AI assistants and why AI infrastructure now faces cloud-service reliability expectations.
Google pushes Gemini onto the World Cup screen
Google is using the 2026 World Cup to place Gemini across Search, lock-screen scores, Maps, Waze and fan image tools.
Apple rents Gemini to rebuild Siri
The new Siri reportedly uses a custom 1.2 trillion-parameter Gemini model under a three-tier routing system that keeps simple work local and heavy reasoning in Google’s cloud.
Apollo and Blackstone line up $36 billion for Anthropic TPUs
A private-credit structure would buy Google TPUs through an SPV and lease them to Anthropic, keeping the debt off Anthropic’s balance sheet.
Google Pledges Water-Positive Data Centers by 2030
Google says it will replenish more water than its data centers consume by 2030, pairing public water reporting with projects across 97 watersheds.
Google Tests an AI Search Opt-Out for UK Publishers
Google is adding Search Console controls that let sites stay out of AI Overviews, AI Mode and AI summaries in Discover, starting with a small UK trial.
Google quietly offers to buy private app code for AI training
A reported Google Play pilot asks developers to sell source code, including dormant projects, while the email itself avoids saying AI.
Alphabet to raise $80B for AI data centers, Berkshire invests $10B
Alphabet plans to raise $80 billion through stock sales to fund AI infrastructure, with Berkshire Hathaway contributing $10 billion.
Google turns Street View into a Genie world model
DeepMind connects 280 billion Street View images to Genie, making real places walkable as stylized 720p worlds for Ultra users.
Google API keys remain usable for 23 minutes after deletion
A security researcher found that revoked Google API keys stay active for an average of 23 minutes, with attackers succeeding over 90% of the time during that window.
Google's Universal Cart lets AI handle your checkout
Google unveiled Universal Cart at I/O 2026, a cross-service shopping cart that uses AI agents to track prices, manage purchases, and complete checkout automatically.
Google Antigravity 2.0 lets 93 AI agents build an OS in 12 hours
Google launched Antigravity 2.0 at I/O 2026, a standalone agent platform that can orchestrate 93 sub-agents to build an operating system in 12 hours.
Google’s Gemini Spark costs $100/month, connects 11 SaaS platforms
Google I/O 2026 introduces Gemini Spark, a 24/7 AI agent with its own Gmail address, priced at $100/month and integrated with 11 SaaS services via MCP.
Gemini 3.5 Flash beats GPT-5.5 in agentic benchmarks, costs just $1.5
Google's Gemini 3.5 Flash outperforms GPT-5.5 on agentic benchmarks while being dramatically cheaper, and becomes the default model across Gemini app and Search.
Google I/O 2026: Gemini 4.0 Isn't the Headline Anymore
Google I/O 2026 kicks off tomorrow, but Gemini's big update is no longer the main event as the company shifts focus to AI agents and hardware.
Google I/O 2026 kicks off tomorrow, Gemini Omni leaks ahead of time
Google's I/O 2026 keynote begins May 19, with leaked details on Gemini Omni, Gemini 3.5, Spark, and Android XR glasses.
Google talks with SpaceX about putting TPUs in orbit
Google’s Project Suncatcher is reportedly moving from research toward launch talks with SpaceX as AI data centers run into power and siting constraints on Earth.
Googlebook replaces Chromebook as Gemini moves into the pointer
Google is reportedly shifting new Chromebook-class laptops to the Googlebook brand, Android base and Gemini-centered interactions such as Magic Pointer.
Google turns Android into a Gemini intelligence layer
At Android Show, Google framed Gemini Intelligence as a system-level layer that can read screens, move across apps and wait for the user’s final approval.
Samsung Brings Gemini Into the Fridge and Rebuilds Bixby
The Bespoke AI Family Hub update adds Gemini vision, AI Food Manager and a renewed Bixby for appliances.
Alphabet's 160% Rally Reprices Its AI Stack
Alphabet's market value has climbed to .8 trillion as investors recast Google from an AI laggard into one of the broadest bets on AI infrastructure, cloud demand and Gemini.
Google Quietly Shuts Down Project Mariner
Project Mariner, Google DeepMind's browser-operating AI agent, has been taken offline as the agent market moves toward structured tools, files and code.
Fitbit Air brings Gemini coaching to a $99 band
Google's screenless Fitbit Air pairs a low upfront price with an optional Gemini-powered health coach, aiming directly at WHOOP's subscription model.
Google tests a middle tier between AI Pro and Ultra
Code strings point to an AI Ultra Lite plan and a Remy upgrade for Gemini Agent, hinting that Google wants both finer subscription pricing and a more persistent assistant.
Google is turning Gemini into Remy, a round-the-clock digital partner
Internal materials describe Remy as a Gemini agent that can act across apps, use personal context and monitor tasks, while Google warns about mistakes and data exposure.
Google Cloud Tops $20 Billion as Gemini Revenue Jumps Eightfold
Alphabet’s Q1 showed cloud revenue, AI demand and Gemini usage turning heavy infrastructure spending into booked revenue.
Google Starts $15 Billion AI Hub in Vizag
Google is building three gigawatt-scale AI data center campuses in Visakhapatnam with AdaniConneX and Airtel, aiming to bring Indian AI inference closer to local users.
Google's classified Pentagon Gemini deal draws internal revolt
More than 600 Google employees opposed a classified Pentagon AI deal that lets Gemini run on military workloads for any lawful government purpose.
Google AI glasses lineup listed, three product tiers take on Meta
Google has laid out a full Android XR glasses lineup spanning three tiers from audio-only to holographic, powered by Gemini and Project Astra, though no release date or pricing has been announced.
Google transforms Vertex AI into an Agent platform
At Google Cloud Next '26, Google renamed Vertex AI to Gemini Enterprise Agent Platform, split its TPU line into training and inference variants, and unveiled the Virgo Network data center architecture, all centered on the agent paradigm.
Google bets $40 billion on Anthropic
Google has committed $10 billion upfront and up to $30 billion more based on performance milestones, alongside a 5-gigawatt compute deal, making it the second-largest external capital source for Anthropic after Amazon.
Google CEO says 75% of new code is now written by AI
Sundar Pichai revealed during a quarterly earnings call that 75% of Google's new code is now AI-generated and human-reviewed, up from 25% 18 months ago. The shift is reshaping developer roles and highlights internal use of rival tools like Claude Code at DeepMind.
EU to Force Google to Give ChatGPT and Claude System-Level Access on Android
The European Commission has initiated a Digital Markets Act procedure to require Google to grant rival AI assistants like ChatGPT and Claude the same system-level access as Gemini on Android, with fines of up to 10% of Alphabet's global annual revenue for non-compliance.
Google splits TPU into two lines: training and inference
At Google Cloud Next, Sundar Pichai unveiled the eighth-generation TPU — not one chip, but two: the TPU 8t for training and the TPU 8i for inference, marking a strategic bet on specialized AI hardware.
Google launches Workspace Intelligence, turning Gmail into an AI agent
At Cloud Next 2026, Google unveiled Workspace Intelligence, a shared semantic layer across Gmail, Docs, Sheets, and more, powered by Gemini to understand your entire work context.
Google Cloud Pledges $750 Million to Help Partners Build AI Agents
Google Cloud announced a $750 million fund on April 22 to help consultancies and system integrators develop and deploy agentic AI, aiming to secure channel influence for its Gemini models.
Google TurboQuant compresses AI inference memory by 6x
Google Research's TurboQuant compresses KV Cache memory for large language models by at least 6x without retraining or accuracy loss, achieving up to 8x speedup on H100 GPUs.
Google adds AI Skills to Chrome, turning prompts into one-click macros
Google's new Skills feature for Gemini in Chrome lets users save frequently used prompts as reusable workflows, accessible with a slash command or a button click. It includes over 50 pre-built templates and supports cross-tab execution, with safety prompts for write operations.