Google bets on Gemini to reinvent the smart home speaker
Google is making a significant move in the smart home market by integrating its Gemini AI model into the Google Home Speaker. This development comes at a time when generative...
Google DeepMind
47 primary articles · 17 secondary mentions
Primary coverage
Unlocking UK house-building with AI-accelerated planning
The National Partnerships for AI has launched an initiative to leverage artificial intelligence in the UK housing sector, aiming to address the pressing need for increased housing supply. This initiative...
Android 17 launches with new multitasking tools as Google expands Gemini features
Google has officially launched Android 17, marking a significant upgrade in its operating system with new multitasking capabilities and enhanced AI features through its Gemini models. This release is particularly...
Google DeepMind is worried about what happens when millions of agents start to interact
Google DeepMind is taking proactive measures to address the potential risks associated with the burgeoning presence of autonomous AI agents. Rohin Shah, the director of AGI safety and alignment research...
Google's Gemini 3.5 Live Translate delivers real-time voice translation across 70+ languages
Google has unveiled its Gemini 3.5 Live Translate feature, significantly expanding its real-time voice translation capabilities in Google Meet. This update allows users to communicate across language barriers more effectively,...
Google Deepmind's Gemma 4 12B squeezes multimodal AI onto a laptop with just 16 GB of RAM
Google DeepMind has unveiled Gemma 4 12B, an open-source multimodal AI model that can operate on laptops equipped with only 16 GB of RAM. This development is significant as it...
AlignAtt4LLM: Fast AlignAtt for Decoder-Only LLMs at IWSLT 2026 Simultaneous Speech Translation Task
Problem This work addresses the gap in simultaneous speech translation systems utilizing decoder-only large language models (LLMs), specifically for the IWSLT 2026 task. Prior implementations of AlignAtt have relied on...
I put Google’s 24/7 AI assistant Gemini Spark to work, and it’s actually pretty useful
Google has launched Gemini Spark, a new AI assistant designed to streamline daily tasks such as summarizing emails and planning local events. This move comes at a time when the...
Gram: Assessing sabotage propensities via automated alignment auditing
Problem This paper addresses the gap in the literature regarding the assessment of AI agents’ propensity to engage in sabotage behaviors, particularly in the context of alignment auditing. The authors...
Google launches a tiny board that runs Gemma 3 locally
Google has introduced the Coral Board, a compact single-board computer designed to run its Gemma 3 AI model locally. This launch, announced at Google I/O, represents a significant step in...
Interpretability-Guided Layer Selection over Subspace Projection: SAEs as Stethoscopes, Not Scalpels, for Raw Task Vector Model Editing
Problem This preprint addresses the limitations of existing model editing techniques for large language models (LLMs), particularly in enhancing domain-specific capabilities without incurring the computational costs or risks of catastrophic...
Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini
Problem This paper presents Gemini Embedding 2, addressing the gap in existing multimodal embedding models that typically struggle to unify video, audio, image, and text modalities into a single representation...
Google Deepmind's AlphaProof Nexus solves decades-old math problems for a few hundred dollars
Google DeepMind’s AlphaProof Nexus has made headlines by autonomously solving nine long-standing Erdős problems, including two that had eluded mathematicians for over half a century. This breakthrough is significant not...
AI models often give the right answers but point to the wrong sources
Problem This paper addresses the phenomenon of “attribution hallucination” in AI models, particularly in the context of large language models (LLMs) like GPT and Gemini. Despite providing correct answers, these...
Deepmind's Hassabis sees humanity "in the foothills of the singularity" while LeCun says current AI isn't intelligent
In a recent discussion among AI thought leaders, Demis Hassabis of DeepMind expressed an optimistic view that humanity is on the brink of a technological singularity, while Yann LeCun, Chief...
We tried Google’s AI glasses and they’re almost there
Google has unveiled a prototype of its Android XR glasses, integrating Gemini-powered features that provide real-time translation, navigation, and contextual information directly in users’ line of sight. This demonstration marks...
Google I/O showed how the path for AI-driven science is shifting
At Google I/O, Demis Hassabis, CEO of Google DeepMind, made headlines by declaring that we are on the brink of a transformative era in AI, referring to it as “the...
We’re launching the Google DeepMind Accelerator program in Asia Pacific to tackle environmental risks
Google DeepMind has announced the launch of its inaugural Accelerator program in the Asia-Pacific region, aimed at addressing pressing environmental challenges exacerbated by climate change. This initiative is timely, as...
Google pairs its Genie world model with Street View to create explorable AI worlds based on real places
Google DeepMind has unveiled an innovative integration of its Genie 3 world model with Street View imagery, allowing users to create walkable, AI-generated environments based on real-world locations. This development...
Google overhauls its AI subscriptions at I/O 2026 with three tiers starting at $10 a month
Google has unveiled a significant overhaul of its AI subscription services during the I/O 2026 event, introducing three distinct pricing tiers ranging from $7.99 to $99.99 per month. This restructuring...
Google’s Genie world model can now simulate real streets with Street View
Google DeepMind has unveiled an exciting enhancement to its Project Genie by integrating Google Street View, enabling the creation of immersive and interactive world simulations. This development is significant as...
With Gemini 3.5 Flash, Google bets its next AI wave on agents, not chatbots
Google has unveiled Gemini 3.5 Flash, a groundbreaking AI model that marks a significant shift from traditional chatbots to more autonomous, agentic capabilities. This launch, presented at the company’s annual...
Gemini 3.5 Flash: frontier intelligence with action
Google DeepMind has unveiled Gemini 3.5 Flash, a significant advancement in AI that enhances the capabilities of intelligent agents and coding workflows. This launch is timely as businesses increasingly seek...
Google’s Gemini Omni turns images, audio, and text into video — and that’s just the start
Google has unveiled its latest innovation, Gemini Omni, a multimodal AI model capable of transforming text, images, and audio into video content through conversational prompts. This launch is significant as...
Google introduces Gemini Spark, a 24/7 agentic assistant with Gmail integration
Google has unveiled Gemini Spark, a groundbreaking agentic personal assistant, during its recent I/O developer conference. This new assistant leverages the capabilities of Gemini’s foundational models and integrates with Gmail,...
Google's I/O announcements: new models, a cloud agent that never sleeps, and a redesigned Gemini app
At Google’s I/O developer conference, the tech giant unveiled a series of significant AI advancements, including the introduction of Gemini 3.5 Flash, a multimodal model named Gemini Omni, and a...
New math benchmark reveals AI models confidently solve problems that have no solution
Problem This paper addresses a significant gap in the evaluation of AI models’ mathematical reasoning capabilities, particularly their ability to discern solvable from unsolvable problems. The authors introduce SOOHAK, a...
Android gets AI agents that book trips, fill forms, and clean up your texts
Google has unveiled Gemini Intelligence, a suite of AI features designed to enhance the Android experience by automating complex tasks. This development is significant as it positions Google to compete...
Alphabet's Isomorphic Labs raises $2.1 billion to scale AI drug discovery toward clinical trials
Isomorphic Labs, the AI-driven drug research firm founded by DeepMind co-founder Demis Hassabis, has successfully secured $2.1 billion in a Series B funding round led by Thrive Capital. This significant...
Everything Google announced at its Android Show, from Googlebooks to vibe-coded widgets
Google recently held its Android Show, showcasing a range of innovative products and features that emphasize its commitment to an AI-first approach. Among the highlights were the new Googlebooks laptops,...
Google brings agentic AI and vibe-coded widgets to Android
Google has unveiled significant updates to its Android platform, introducing agentic AI capabilities and innovative vibe-coded widgets. This move is part of the tech giant’s ongoing effort to enhance user...
Google adds Gemini-powered Dictation to Gboard, which could be bad news for dictation startups
Google has unveiled a new dictation feature powered by its Gemini AI model, integrating it into the Gboard keyboard for Samsung Galaxy and Google Pixel devices. This move is significant...
Classifier Context Rot: Monitor Performance Degrades with Context Length
Problem This preprint addresses a significant gap in the evaluation of language models (LMs) for monitoring coding agents, particularly in the context of long transcripts exceeding 500K tokens. Existing benchmarks...
Google DeepMind, Microsoft and xAI sign agreements with US government-backed AI safety institute to test... - Moneycontrol.com
In a significant move for AI safety, Google DeepMind, Microsoft, and xAI have entered agreements with a U.S. government-backed AI safety institute. This collaboration aims to rigorously test AI systems...
PSK at SemEval-2026 Task 9: Multilingual Polarization Detection Using Ensemble Gemma Models with Synthetic Data Augmentation
Problem This paper addresses the challenge of multilingual polarization detection, a binary classification task across 22 languages, as part of SemEval-2026 Task 9. The authors identify a gap in existing...
US Government Expands AI Testing Agreements to Google DeepMind, Microsoft, and xAI - H2S Media
The U.S. government has broadened its AI testing agreements to include major players like Google DeepMind, Microsoft, and xAI, a company founded by Elon Musk. This expansion is significant as...
Google DeepMind, Microsoft and xAI Sign Agreements for US National Security AI Testing - Technobezz
In a significant move for the intersection of artificial intelligence and national security, Google DeepMind, Microsoft, and xAI have entered into agreements to collaborate on AI testing for U.S. defense...
Google DeepMind Employees Launch Unionization Effort, Citing Military AI Deals
In a significant move for labor rights within the tech sector, Google DeepMind employees in the U.K. have initiated a unionization effort, seeking recognition from management for the Communication Workers...
UnAC: Adaptive Visual Prompting with Abstraction and Stepwise Checking for Complex Multimodal Reasoning
Problem This paper addresses the limitations of large multimodal models (LMMs) in performing complex reasoning tasks that require multi-step inference over visual data. Despite advancements in visual perception capabilities, existing...
CAISI Signs Agreements Regarding Frontier AI National Security Testing With Google DeepMind, Microsoft and xAI - National Institute of Standards and Technology (.gov)
The Center for AI Safety and Innovation (CAISI) has announced significant agreements with tech giants Google DeepMind, Microsoft, and xAI to advance national security testing for frontier artificial intelligence. This...
news
Google DeepMind
DeepMind CEO Praises DeepSeek, Calls Hype Exaggerated - Let's Data Science
DeepMind’s CEO has recently expressed a mixed sentiment regarding the AI tool DeepSeek, praising its capabilities while simultaneously cautioning against the exaggerated hype surrounding it. This statement comes at a...
How to Analyze Customer Sentiment with Google Gemini
Google has introduced Gemini, a generative AI tool designed to help businesses analyze customer sentiment more efficiently. This development comes at a time when companies are inundated with unstructured feedback...
Google’s Gemini AI assistant is hitting the road in millions of vehicles
Google is set to launch its Gemini AI assistant in millions of vehicles equipped with Google built-in, marking a pivotal upgrade from the existing Google Assistant. This initiative underscores Google’s...
Beyond Semantics: Measuring Fine-Grained Emotion Preservation in Small Language Model-Based Machine Translation
Problem This preprint addresses the gap in the literature regarding the preservation of fine-grained emotional nuances in Machine Translation (MT) using Small Language Models (SLMs). While existing MT systems prioritize...
How Generative AI Disrupts Search: An Empirical Study of Google Search, Gemini, and AI Overviews
Problem This preprint addresses the gap in understanding how generative AI alters the landscape of web search, particularly in comparison to traditional search engines. The authors investigate the differences in...
Enabling a new model for healthcare with AI co-clinician
A groundbreaking initiative by Google DeepMind aims to revolutionize healthcare by introducing an AI co-clinician designed to assist medical professionals in diagnosing and treating patients. This development comes at a...
news
Google DeepMind
More Gemini features are coming to Google TV
Google TV has recently integrated additional features from Gemini, Google’s AI-driven multimedia tool, enhancing user experience with innovative capabilities. Notably, tools like Nano Banana and Veo allow users to transform...