Skip to content

Organizations

Google / DeepMind · Latest

Models, research and announcements

18 selected reportsIn the last 30 days 16 reportsIndexed 94 reports

RSS
updated

Selected: Google / DeepMind

TodayOct 31SatItems 1–18
  1. The Decoder

    Reddit CEO Steve Huffman Blames Google’s AI Mode for Loss of Over One Million Weekly Search Referrals

    Reddit’s shares fell more than 25 % after the company reported that it lost over one million Google search referrals per week in October, citing Google’s new AI Mode and AI Overviews as the cause. Huffman said AI Mode does not drive traffic to Reddit or to any other site, describing it as a fundamental shift in how Google operates. Reddit claims user demand for its platform has not changed, but Google has stopped sending users there. In response, Reddit is testing login walls outside the U.S. to curb scraping and reduce its dependence on the open web, while continuing data‑licensing talks with AI providers. The incident highlights a structural risk for any business that relies on Google for traffic.

    Read source the-decoder.com
Oct 29Thu
  1. Google DeepMindFirst-party

    Google DeepMind announces Gemini 4 Carbon, its strongest AI model with state-of-the-art cybersecurity and long-horizon capabilities

    Google DeepMind has unveiled Gemini 4 Carbon, its most powerful AI model to date, ranking #1 on the Artificial Analysis Intelligence Index v5.0 with a score of 59. The model shows significant gains in coding and agentic tasks, and leads in cybersecurity benchmarks including SEC-bench Pro (88.2%), ExploitGym (54.8%), and CWE-bench (71.0). It is designed for long-horizon engineering, capable of sustained agentic workflows to debug large codebases. Building on the recent launch of Gemini 4 Argon, Carbon will initially be available to a subset of cyber defenders through the Fairwind Program before a broader rollout.

    Read source deepmind.google
Oct 28Wed
  1. Google ResearchFirst-party

    Google Research introduces sleep framework for LLMs to retain knowledge in continual learning

    Google Research proposes a sleep framework for continual learning in large language models, enabling them to retain knowledge beyond the context window. The framework operates in alternating active and offline sleep phases, during which the model consolidates short-term context into stable long-term parameters using Mixture-of-Experts memory blocks. This approach allows LLMs to continuously acquire new facts, retain long-context knowledge up to 10M tokens, and improve complex reasoning without altering the core architecture.

    Read source research.google
  2. Google DeepMindFirst-party

    DeepMind Unveils Genie 4, a Real‑Time World Model with Native Audio Generation

    On October 27, 2026 DeepMind announced Genie 4, a multimodal world simulation model that produces photorealistic visuals, fluid physics, and native audio—including music, ambient sounds, and interactive dialogue—simultaneously. Built on the Gemini family, Genie 4 uses Gemini Omni for realistic appearance and behavior, while other Gemini models supply the reasoning needed to keep simulations coherent. Users can steer scenes in real time with natural‑language prompts that alter weather, camera angles, or introduce new objects, with Gemini adapting the context. The model is currently in a research preview for trusted testers, with plans to roll out to Google AI Ultra subscribers and Gemini API partners. Genie 4’s ability to generate realistic physics, human motion, and sound in real time represents a significant step toward safe, infinitely variable simulations for robotics training, AI agents, and new interactive experiences.

    Read source deepmind.google
Oct 22Thu
  2. The Decoder

    Deepmind founder Hassabis shortens AGI timeline to 2-3 years

    Deepmind founder Demis Hassabis now estimates artificial general intelligence could arrive within two to three years, shortening his previous forecast of three to five years. He defines AGI as a system matching the full cognitive range of the human brain and uses the "Einstein test" as a benchmark, asking whether a system trained only on pre-1901 knowledge could reproduce Einstein's 1905 papers.

    Read source the-decoder.com
  3. Google ResearchFirst-party

    Google Research develops AI model for gestational age assessment using blind sweep ultrasonography

    Google Research published a study in JAMA Network Open demonstrating that an AI model can reliably estimate gestational age from blind sweep ultrasounds performed by novice operators in real clinics. The model maintained accurate performance across diverse settings including Northwestern Medicine in Chicago and a field site in Nairobi, Kenya, with minimal fine-tuning. The research addresses the global shortage of ultrasound specialists by enabling frontline health workers to perform scans with minimal training.

    Read source research.google
Oct 21Wed
  1. NVIDIA BlogFirst-party

    Google Cloud and NVIDIA launch NVIDIA Vera Rubin infrastructure in general availability

    Google Cloud and NVIDIA announced general availability of NVIDIA Vera Rubin infrastructure next month, bringing an integrated platform for training, inference, and large-scale reinforcement learning to cloud customers. The A5X instances combine NVIDIA Vera Rubin NVL72, ConnectX-9 SuperNICs, and Google Cloud's fourth-generation liquid-cooling technology to support demanding agentic workloads at scale.

    Read source blogs.nvidia.com
Oct 18Sun
  1. The Decoder

    German Courts Hold Google Liable for AI‑Generated Search Summaries

    In May 2026 the Munich Regional Court ruled that Google’s AI overviews are the company’s own content, making Google directly liable for false statements. A later Aachen court applied the same logic, issuing an injunction against further dissemination of fabricated claims about a physician and threatening fines of up to €250,000 per violation. The rulings underscore that AI‑generated answers are treated as original content, not merely neutral search results.

    Read source the-decoder.com
Oct 15Thu
  2. Google ResearchFirst-party

    Google Research launches Deep Review for technical document analysis

    Google Research has released Deep Review, an AI system designed to provide rigorous feedback and technical verification for complex documents. Initially developed for computer science papers, it now supports multiple domains and offers a scalable API. The system aims to address the review bottleneck in academic publishing, with over 74,000 submissions to major AI conferences in 2026.

    Read source research.google
Oct 14Wed
  1. Google DeepMindFirst-party

    Google DeepMind Launches Gemini 4 Argon, a Frontier Model for Complex Engineering and Business Workflows

    Google DeepMind has released Gemini 4 Argon, a new frontier AI model now available globally via API, AI Studio, and the Gemini platform. Argon excels at multi‑step tasks in coding, finance, legal, cybersecurity, and science, topping the Vals Index and outperforming other frontier models. It is priced at $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached inputs. Early adopters at Google and enterprises are using Argon to refactor massive codebases, optimize business processes, patch security flaws, and accelerate research in quantum computing and life sciences.

    Read source deepmind.google
Oct 8Thu
  1. Google ResearchFirst-party

    Google Research Explores AI's Impact on Professional Judgment

    A three-month field experiment by Google Research found that AI tools can significantly boost patent drafting scores among lawyers, but the benefits may not persist over time. The study reveals the complexities of AI's role in enhancing or eroding expertise.

    Read source research.google
Oct 6Tue
  2. Google ResearchFirst-party

    Google Report on Agent Privacy and Security: Contextual Integrity Framework

    Google's report identifies foundational privacy and security challenges in autonomous agents, emphasizing the need for contextual integrity in their actions. The report, based on the theory of Contextual Integrity, outlines the complexities of agents processing unstructured data and taking probabilistic actions, highlighting the importance of understanding context and appropriateness in their operations.

    Read source research.google
Oct 2Fri
Sep 30Wed
  1. Google ResearchFirst-party

    Google Research's Diffusion Controller Simplifies AI Image Generation

    Google Research introduces the Diffusion Controller framework, redefining the AI image generation process as a continuous control problem. The framework, with its lightweight steering damper network, achieves precise image control without altering the core model weights, outperforming industry standards.

    Read source research.google
Sep 25Fri
  1. Google DeepMindFirst-party

    Google DeepMind launches Gemini 3.8 Live with Live Avatar

    Google DeepMind has released Gemini 3.8 Live with Live Avatar, a feature that couples low‑latency video streaming with Gemini’s live dialogue models to create a near real‑time visual presence. The system delivers precise lip‑syncing, natural facial expressions, and fluid turn‑taking across 97 languages, while still enabling asynchronous tool calls in the background. Enterprises can deploy brand‑customized avatars for customer service, interactive walkthroughs, and other virtual offerings, all within the Gemini Enterprise platform.

    Read source deepmind.google