From Google DeepMind · 40 stories
Meta Introduces Hybrid Asset Classification for Privacy-Aware Infrastructure
Meta has unveiled a hybrid asset classification strategy using large language models (LLMs) to handle ambiguous data in privacy-aware infrastructure while maintaining deterministic rules for enforcement. This method addresses the complexities of AI-native products with varied data inputs, ensuring compliance and effective data governance. It is a response to the challenges posed by the increasing speed and scale of AI innovations, and the approach aims to better manage privacy controls for evolving AI products.
Google Unveils Gemini 4 Argon AI Model, Phased Rollout Begins with Cyber Defenders
Google has announced Gemini 4 Argon, its new flagship AI model, which demonstrates advanced capabilities in coding, reasoning, and cybersecurity defense. The model features a 1M output token limit and surpasses top models from OpenAI and Anthropic in several benchmarks. Access to Gemini 4 Argon is being rolled out in phases, starting with trusted cyber defenders and internal Google teams, before wider availability to paid API customers, AI Ultra subscribers, developers, and enterprises.
Google Expands Gemini Enterprise Agent Platform with Remote MCP Server
Google has enhanced its Gemini Enterprise Agent Platform by introducing a remote Managed Control Plane (MCP) server. This update allows developers to securely connect external AI agents with Google Cloud resources, facilitating agent development across various IDEs. The enhancements address developer feedback on building more efficient, production-ready AI agents.
Google releases Gemini 3.8 Flash AI model with improved benchmarks
Google has released Gemini 3.8 Flash, the third Flash update in three months, which shows improved performance in benchmarks compared to its predecessor. This model is designed for software engineering, autonomous agents, and complex enterprise workflows, and is available with an introductory pricing offer.
Google DeepMind and Google Research Release WeatherNext 3 AI Weather Model
Google DeepMind and Google Research have released WeatherNext 3, an AI model for global weather forecasting. The model provides hourly predictions with higher resolution, leveraging real-time satellite observations, and has demonstrated higher accuracy than other leading AI and traditional forecasting systems in independent evaluations.
Google may soon allow customization of Gemini voice parameters
Google's upcoming Gemini update could enable users to customize voice characteristics such as Energy, Formality, Speed, and Warmth. This feature will enhance user interaction with Gemini across various applications, likely improving personalization options for voice interactions.
Google DeepMind Introduces Gemini 3.8 Live Models for Voice Agents
Google DeepMind launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two new AI models designed to improve near real-time reasoning for voice agents. These models aim to make AI conversations more intuitive and intelligent, with Gemini 3.8 Live for scalable applications and Gemini 3.8 Live Extended Thinking for high-complexity tasks requiring multi-step reasoning.
Google Launches Gemini 3.5 Flash Cyber for Efficient Vulnerability Management
Google DeepMind has launched Gemini 3.5 Flash Cyber, a cybersecurity AI model to detect and patch vulnerabilities efficiently. Initially available to governments and trusted partners via CodeMender, this model provides a cost-effective alternative to larger AI security models like Anthropic's Mythos. Its introduction is part of a broader release including Gemini 3.6 Flash and 3.5 Flash-Lite, demonstrating Google's continued commitment to cybersecurity innovation.
Google Introduces Gemini Omni for AI-Powered Video Editing
Google launched the Gemini Omni model to enhance video editing with AI, allowing users to edit videos using text, image, and audio inputs. The feature debuts as the Video Remix tool in Google Photos and the Gemini Omni Flash model in apps like YouTube Shorts. This development simplifies video editing for both individual users and businesses, integrating AI further into Google's ecosystem.
Google Launches Faster AI Models for Image and Video: Nano Banana 2 Lite and Gemini Omni Flash
Google announced the launch of Nano Banana 2 Lite and Gemini Omni Flash, models designed to enhance multimedia processing efficiency. Nano Banana 2 Lite focuses on low-cost, fast image generation, while Omni Flash offers advanced video generation and editing. Available across Google platforms, these models aim to streamline creative workflows for developers and businesses using AI-generated content.
Google DeepMind open-sources WeatherNext AI for 15-day cyclone forecasts
Google DeepMind and Google Research have open-sourced the WeatherNext AI model, which provides 15-day forecasts for tropical cyclone track and intensity. The model uses Functional Generative Networks (FGNs) to generate forecasts in under a minute, offering an average of one day more lead time than existing models. This release allows other scientists to utilize and build upon the model, potentially improving early warning systems for hurricanes and typhoons.
Google commits $40 million to advance the Genesis Mission for scientific research
Google has pledged $40 million in AI tokens and cloud credits to support the Genesis Mission, which aims to enhance American scientific discovery using AI. This funding will provide access to advanced AI tools such as AlphaFold 3 and Gemini, aimed at DOE laboratories, significantly expanding capabilities in various scientific disciplines.
Google DeepMind announces $10M funding for multi-agent AI safety research
Google DeepMind has announced a new funding initiative of up to $10 million for research on multi-agent AI safety. This funding aims to understand and manage the risks associated with interactions among AI agents as they become more widespread, which is critical for ensuring safety and predictability in AI systems.
Google Gemini 3.5 Live Translate Offers Real-Time Voice Translation
Google has launched Gemini 3.5 Live Translate, enhancing speech-to-speech translation for over 70 languages. The model allows for continuous translation without pauses, significantly improving real-time multilingual communication.
DeepMind's SL2T Model Enables American Sign Language Input on Pixel 11 Devices
DeepMind has launched its new sign-language-to-text (SL2T) model, integrated into Gboard and Live Transcribe on Pixel 11 devices. This technology allows users to transcribe American Sign Language (ASL) into English text, providing a new method for communication for deaf and hard of hearing individuals.
Gemma 4 12B Boosts Multimodal AI Processing on Laptops
Google DeepMind introduced Gemma 4 12B, a new encoder-free multimodal AI model, enabling advanced processing on laptops with minimal memory. Gemma 4's architecture eliminates multimodal encoders, creating efficient audio and visual input processing. Collaboration with Cerebras and Hugging Face enhances real-time speech-to-speech capabilities, improving applications like voice assistants.
SynthID Bio Introduces Watermarking for AI-Generated Biological Designs
SynthID Bio has launched a watermarking approach for AI-generated biological designs to strengthen biosecurity and information integrity. This technology embeds a verification layer into biological designs, helping to track provenance and streamline DNA synthesis screening.
Google Introduces Private AI Compute Architecture for Server-Side Memory
Google announced an update to its Private AI Compute architecture, introducing persistent, server-side memory with on-device privacy standards. This allows AI assistants to maintain long-term continuity across devices while keeping user data encrypted and accessible only by the user's devices.
Gemini Omni 1.1 Flash Adds Creative Controls and Generative Video Capabilities for Developers
Google has updated Gemini Omni 1.1 Flash with new creative controls and generative video capabilities, making it production-ready for professional use via the Gemini API in Google AI Studio. These updates allow developers to extend video scenes, specify first and last frames for transitions, and draft videos more efficiently at lower resolutions and costs, enhancing control and iteration speed for generative video workflows.
Google Launches Antigravity 2.0 Desktop Application
Google has released Antigravity 2.0, a standalone desktop application for macOS, Linux, and Windows, enabling interaction with AI agents. This update improves agent functionality with features like dynamic subagents, asynchronous task management, and project-based organization.
Google DeepMind and Isomorphic Labs Unveil Bioresilience Strategy
Google DeepMind and Isomorphic Labs have announced their approach to bioresilience, focusing on preventing AI misuse while leveraging AI to enhance global health responses. In the past year, over 15 partnerships have been established to address biosecurity challenges through advanced AI technologies.
ATL Saathi utilizes Gemini AI to enhance innovation in Indian schools
ATL Saathi introduces AI-powered tools to 100 pilot schools in India, transforming traditional tinkering labs. The Gemini model supports teachers with curriculum-aligned resources, aiming to reduce administrative burdens and foster creativity among students.
Google DeepMind partners with A24 to enhance film production through research
Google DeepMind and A24 have formed a partnership aimed at combining advanced AI research with filmmaking. This collaboration will provide filmmakers with innovative tools while allowing DeepMind to receive feedback from industry artists, ultimately enhancing creative workflows.
Gemini 3.5 Flash Integrates Built-In Computer Use Tool
Gemini 3.5 Flash now includes a built-in computer use feature, enhancing automation and testing tasks for developers. This integration allows for improved functionality in various environments and addresses prompt injection risks through training and enterprise safeguards.
UK government partners with AI for faster house-building planning
The UK government is developing an AI tool with Google DeepMind to reduce householder application processing times by 50%. This prototype aims to alleviate bottlenecks in local planning authorities, facilitating the construction of 1.5 million new homes by 2029.
Google's AI Control Roadmap Enhances Security for AI Systems
Google introduces its AI Control Roadmap to safeguard AI agents against risks from imperfect alignment. This framework aims to ensure security and reliability in AI deployment, treating AI agents as potential insider threats.
DiffusionGemma Model Achieves 4x Faster Text Generation
DiffusionGemma, a new 26B Mixture of Experts model, enables text generation up to 4x faster than traditional autoregressive models. Targeted at researchers, it aims to improve speed for interactive workflows despite lower output quality than its predecessor, Gemma 4.
Google DeepMind Launches AI Accelerator Programs in Singapore and Europe
Google DeepMind has announced AI-driven initiatives in Singapore and Europe. In Singapore, the focus is on healthcare, education, and workforce development as part of a national partnership, aiming to contribute S$3.3 billion to the economy by 2040. In Europe, a three-month robotics accelerator will support startups in developing AI applications in the physical world.
AI Augments Learning in Sierra Leone, Showing Significant Student Gains
A trial in Sierra Leone demonstrates that AI, through the Guided Learning approach, enhances student understanding without replacing teachers. Students improved their math scores significantly, showcasing the potential of AI in educational settings.
Co-Scientist accelerates genetic research to reverse cellular aging
Biologists Omar Abudayyeh and Jonathan Gootenberg utilize Co-Scientist to identify genetic pathways that could reverse cellular aging. The tool significantly reduces both the lead generation process and data analysis time, facilitating more efficient aging research.
Project Genie adds Google Street View capabilities for virtual environments
Google's Project Genie now incorporates Google Street View imagery, enhancing its ability to create interactive virtual environments. This update allows users to generate imaginative worlds based on real-world locations, expanding the simulation possibilities for AI and users alike.
Google Launches Gemini for Science to Enhance Scientific Research
Google introduced Gemini for Science, a suite of AI tools aimed at enhancing scientific research. This collection includes prototypes for hypothesis generation, computational discovery, and literature insights, facilitating faster and more efficient scientific exploration.
Expanded tools for content transparency in generative media
New content transparency and verification tools have been launched across various platforms. This enhancement aims to help users understand the origin and authenticity of generative media content.
Researcher uses AI to identify targets in infectious disease study
Professor Clare Bryant utilized the AI tool Co-Scientist to identify molecular switches linked to infectious diseases. This approach has accelerated her research timeline from two to three years to potentially six months.
Calico Life Sciences employs AI for novel aging research hypotheses
Calico Life Sciences is using its AI tool, Co-Scientist, to analyze biological data on aging. By improving the identification of valuable hypotheses, this approach facilitates advancements in understanding the integrated stress response and its implications for health and disease.
University of Edinburgh Uses Co-Scientist to Advance Liver Disease Research
The University of Edinburgh is leveraging Co-Scientist to analyze literature on metabolic dysfunction-associated steatohepatitis (MASH) for better treatment hypotheses. The platform identified the NLRP3 inflammasome as a key factor linking inflammation and metabolism, potentially guiding dual-therapy development.
Co-Scientist bridges labs in innovative ALS research collaboration
Ritu Raman and Ryan Flynn collaborate to investigate ALS using Co-Scientist, merging their unique methodologies. This integration aims to explore RNA-based mechanisms for potential therapeutic approaches to ALS.
AI aids discovery of repurposed drugs to combat liver fibrosis
Research led by Gary Peltz at Stanford utilizes Co-Scientist to identify existing drugs for liver fibrosis treatment. The AI selected promising candidates, including vorinostat, which showed significant efficacy in lab tests.
Google Flow Music integrates Lyria 3.5 for improved music generation
Google has launched Lyria 3.5 in Google Flow Music, enhancing musicality, lyrics, and vocal quality in generated music. This update provides users with more creative control over tempo and duration, aiming to improve the quality of AI-generated songs.
Google DeepMind Reflects on 15 Years of AI Research Driven by Games
Google DeepMind reviewed its 15-year history of using games like Atari, Go, and StarCraft II to advance AI research, highlighting key breakthroughs such as Deep Q-Networks and AlphaGo. This retrospective emphasizes how game environments have served as critical testbeds for developing more capable and general AI systems.