AI Lab
Official Google DeepMind announcements — model releases, product launches and research, each one summarized with every source covering it, by AIBriefs. RSS
Google Research introduces Science One, a framework for autonomous scientific research that uses Chain-of-Evidence to generate verifiable results.
Google's Gemini Spark assistant now integrates with Chrome for web browsing capabilities, enabling real-time information retrieval and enhanced responses.
Google's open-source TPU microbenchmark suite provides granular performance metrics across five components: Network, Compute, HBM, Host Transfer, and Attention. Developers can use these to establish a Roofline model for TPU performance analysis.
The new model powers Google Flow Music with more expressive vocals, richer arrangements, and enhanced creative controls including BPM adjustment. It represents the latest version of DeepMind's music generation model.
The Gemini app for macOS now supports natural language voice input for generating transcriptions, edits, and summaries.
Google.org launched three new AI tools and training programs to help nonprofits leverage Google AI, announced July 28.
Gemini Robotics 2 brings whole-body intelligence, multi-robot collaboration, and advanced dexterity to humanoids, based on Gemini. The companion Gemini Robotics ER 2 model enhances video understanding and task orchestration.
Google Search introduces AI-powered tools to help users craft menus, design tablescapes, and manage party-planning tasks. The features are available in the Google app and on desktop.
Google has signed the EU AI Act Code of Practice on transparency, committing to label AI-generated content. The move reinforces Google's commitment to responsible AI development in Europe, following the EU's regulatory framework for AI.
Google's ATLAS report reveals Gemini users are augmenting their tasks, not automating them away. OpenAI's research finds ChatGPT users take on new roles across job boundaries, reshaping work without replacing jobs.
Google Research introduces SymptomAI, a conversational AI agent for everyday symptom assessment. The system incorporates responsible AI principles and leverages natural language processing for health-related conversations.
Google researchers developed a method for quantum computers to learn from errors, aiming to improve fault tolerance. The approach uses machine intelligence to adaptively correct noise in quantum circuits.
Google's Gemini Live now works with your phone's camera to analyze what you see in real time. This guide walks through setup and usage tips.
Gemini 3.6 Flash and 3.5 Flash-Lite are GA with 1M token context, 64k output, thinking, and Computer Use. Temperature, top_p, and top_k are deprecated. Pricing is lower than prior generations.
Google published 13 tips for college students to use AI tools like Gemini and AI Mode for summer productivity. Tips cover preparation for grad school, internships, and fall courses. Also includes advice on using YouTube and other Google products.
Tunix is a new JAX-native post-training library that eliminates TPU idling bottlenecks when training multi-turn, tool-using LLM reasoning agents. It maximizes hardware throughput by combining highly concurrent, asynchronous rollouts with a decoupled producer-consumer pipeline.
Google's Gemini can help design brand identity, conduct market research, and automate logistics for side businesses. The post outlines five practical ways to leverage Gemini's capabilities for launching a side hustle.
Ray 2.55 introduces official, first-class support for Google Cloud TPUs, enabling distributed Python workloads via familiar Ray APIs. Multi-host TPU slices must be kept together, handled by Ray's placement groups.
Google partners with University of Waterloo's Futures Lab to help students build real-world AI tools. The initiative aims to prepare students for AI-driven jobs of the future.
Google Vids now allows users to generate and edit videos via natural language prompts using Gemini Omni and create digital avatars from a selfie and voice recording. The Gemini Omni Flash model currently holds the #1 spot on the Artificial Analysis Text-to-Video and Image-to-Video leaderboards.
Google's AI note-taking app NotebookLM is now called Gemini Notebook, though it remains a standalone product. The app will integrate more deeply with Google services and Gemini Search. The rename builds on the former Project Tailaway name revealed last year.
Google is partnering with the Georgia Public Library Service to offer no-cost career and AI training across the state. The program will be delivered through the public library system.
Users can now link and interact with services like Gmail and Calendar directly in AI Mode. The feature expands AI Mode from answer-based queries to performing tasks across linked apps.
The blog outlines their joint approach to using AI for biological threat detection and pandemic preparedness. It covers risk assessment frameworks and model development.
Google and Screwfix have partnered to provide AI tools for UK tradespeople, focusing on administrative tasks and business growth. The collaboration aims to help tradespeople save time on paperwork and improve productivity using Google's AI technologies.
Google advocates treating prompts as build artifacts via modular 'skill files' compiled by a transpiler to enforce static constraints. The approach targets scaling bottlenecks and runtime errors from monolithic system prompts.
Conductor has evolved from a Gemini CLI extension into a portable plugin, bringing conversational Spec-Driven Development to the Antigravity ecosystem. Developers can chat naturally with their AI assistant instead of relying on strict command sequences.
Google Cloud integrates Parallel Web Systems' search infrastructure as a web grounding provider for the Gemini Enterprise Agent Platform. This allows developers to anchor AI agents in verifiable, real-time web results, expanding choice in grounding sources.
A study shows diffusion model creativity arises from neural networks learning a smoothed score function, driving interpolation between training data points. The work, presented at ICLR 2026, mathematically explains how models generate novel data rather than memorizing the training set.
The Metropolitan Museum of Art and Google Arts & Culture unveiled two new generative AI initiatives to celebrate 15 years of partnership. The projects aim to enhance visitor engagement and explore cultural heritage through AI-powered experiences.
Google is rolling out Gemini AI features in Chrome to UK users starting July 14, 2026. The expansion integrates Gemini into the Chrome browser for UK users.
Gemini's local language support drives adoption across Southeast Asia's mobile-first population. The report highlights daily usage patterns in markets like Indonesia and Thailand.
ATL Saathi, powered by Gemini, helps educators in robotics labs across India. The tool was developed in partnership with the Atal Innovation Mission to support STEM education through AI-driven assistance.
At Google I/O Connect India, Google showcased the lightweight Gemma 4 E2B model running natively on the Tensor TPU of the new Pixel 10 family. The model enables completely offline multimodal features, emphasizing 100% private on-device AI.
TabFM is a foundation model that performs classification and regression on tabular data without requiring dataset-specific training or hyperparameter tuning. It uses in-context learning to make predictions on unseen tables in a single forward pass.
Google's blog explains steps to create study notebooks using Gemini, including organizing notes and generating summaries. The feature is available in the Gemini app.
DeepMind podcast explores interpretability, the science of reverse engineering how neural networks learn and think. Host Hannah Fry discusses with researcher Neel Nanda how a model's chain of thought acts like a scratch pad, offering a window into its reasoning.
New features help consumers understand when ads use AI-generated content. Advertisers get simple disclosure tools as part of Google's transparency push.
AlphaEvolve, developed by Google DeepMind, is now widely available on Google Cloud to help solve customers' hardest problems. The rollout targets the most challenging issues on the platform.
LiteRT.js is a new JavaScript library for running ML models directly in the browser, part of Google's LiteRT family. It uses WebAssembly and GPU acceleration for high performance, extending Google's edge AI runtime to the web.
Google Developer Experts built a real-time AI race coach at Sonoma Raceway using Antigravity and Gemini, securing a 0.1-second advantage per lap. Antigravity serves as a domain-bridging engine, grounding AI in physics and real-time verification to close the AI trust gap for high-stakes decisions.
Google outlines three ways it works with governments and organizations on crisis resilience using AI breakthroughs. The initiatives include predictive models and real-time data analysis for disaster response.
Managed Agents now support background tasks, remote MCP servers, and custom function calling within the Gemini API. A free tier is now available to try the service, along with new cost controls and scheduled triggers.
Google's elastic training via Pathways lets multi-node TPU training survive individual machine failure, recovering in seconds. The capability, part of MaxText, was demonstrated by terminating a TPU mid-training with full recovery.
B3, a Brazilian exchange, selected Android Enterprise with Samsung devices to securely scale mobile device management and AI capabilities for its management workforce. The deployment aims to enhance productivity while maintaining security.
The summit brought together NYC educators and industry leaders at Google's offices to discuss AI integration in education. Specific outcomes were not disclosed.