Google’s Gemini transparency cut leaves enterprise developers ‘debugging blind’
Google’s gemini transparency cut leaves enterprise developers ‘debugging blind’
📝 要約
Why is google hiding gemini's reasoning traces?
Unlock the other 99% of your data – now ready for AI
Unlock the other 99% of your data – now ready for ai
📝 要約
For decades, companies of all sizes have recognized that the data available to them holds significant value, for improving user and customer experienc...
It’s pretty easy to get DeepSeek to talk dirty
It’s pretty easy to get deepseek to talk dirty
📝 要約
Ai companions like replika are designed to engage in intimate exchanges, but people use general-purpose chatbots for sex talk too, despite their stric...
The OpenAI Files: Ex-staff claim profit greed betraying AI safety
The openai files: ex-staff claim profit greed betraying ai safety
📝 要約
‘the openai files’ report, assembling voices of concerned ex-staff, claims the world’s most prominent ai lab is betraying safety for profit.
A timeline of the US semiconductor market in 2025
A timeline of the us semiconductor market in 2025
📝 要約
It’s already been a tumultuous year for the u.
GenLayer launches a new method to incentivize people to market your brand using AI and blockchain
Genlayer ローンチ a new method to incentivize people to market your brand using ai and blockchain
📝 要約
With applications like rally already live in beta, genlayer presents a new category of intelligent blockchain infrastructure.
Apple hints at AI integration in chip design process
Apple hints at ai integration in chip design process
📝 要約
Apple is beginning to use generative 人工知能 to help design the chips that power its devices.
Announcing our 2025 VB Transform Innovation Showcase finalists
Announcing our 2025 vb transform innovation showcase finalists
📝 要約
Seven companies will be sharing their latest ai innovations from the main stage at vb transform in sf on june 25.
OpenAI open sourced a new Customer Service Agent framework — learn more about its growing enterprise strategy
Openai open sourced a new customer service agent framework — learn more about its growing enterprise strategy
📝 要約
By offering transparent tooling and clear implementation examples, openai is pushing agentic systems out of the lab and into everyday use.
Announcing the 2025 finalists for VentureBeat Women in AI Awards
Announcing the 2025 finalists for venturebeat women in ai awards
📝 要約
Announcing the finalists for the 2025 women in ai awards.
Here are the 24 US AI startups that have raised $100M or more in 2025
Here are the 24 us ai startups that have raised $100m or more in 2025
📝 要約
Last year was monumental for the ai industry in the u.
‘Kid-pilled’ Sam Altman ‘constantly’ asked ChatGPT questions about his newborn
‘kid-pilled’ sam altman ‘constantly’ asked chatgpt questions about his newborn
📝 要約
Sam altman, who is both the father of a 3-month-old and ceo of openai, hopped on openai's new podcast today to talk about how his company is impacting...
Six-month-old, solo-owned vibe coder Base44 sells to Wix for $80M cash
Six-month-old, solo-owned vibe coder base44 sells to wix for $80m cash
📝 要約
In base44's six months as a stand-alone company, it reportedly grew to 250,000 users and was making nearly $200,000 in monthly profits.
Multiplier, founded by ex-Stripe exec, nabs $27.5M to fuel AI-powered accounting roll-ups
Multiplier, founded by ex-stripe exec, nabs $27.5m to fuel ai-powered accounting roll-ups
📝 要約
Multiplier raised series a and seed 資金調達 from lightspeed and ribbit capital.
Midjourney launches its first AI video generation model, V1
Midjourney ローンチ its first ai video generation モデル, v1
📝 要約
Midjourney launched its much anticipated ai video generation モデル, v1, on wednesday and laid out a roadmap for its future releases.
OpenAI can rehabilitate AI models that develop a “bad boy persona”
Openai can rehabilitate ai models that develop a “bad boy persona”
📝 要約
A new 論文 from openai has shown why a little bit of bad 学習 can make ai models go rogue—but also demonstrates that this problem is generally pr...
OpenAI drops Scale AI as a data provider following Meta deal
Openai drops scale ai as a data provider following meta deal
📝 要約
Openai says it's cutting ties with scale ai following its deal with meta, raising questions about the data provider's business.
xAI is facing a lawsuit for operating over 400 MW of gas turbines without permits
Xai is facing a lawsuit for operating over 400 mw of gas turbines without permits
📝 要約
The colossus data center was allegedly operating as many as 35 gas turbines without pollution controls in a region that already had poor air quality.
OpenAI found features in AI models that correspond to different ‘personas’
Openai found features in ai models that correspond to different ‘personas’
📝 要約
By looking at an ai モデル's internal representations — the numbers that dictate how an ai モデル responds, which often seem completely incoherent to hu...
The ‘OpenAI Files’ push for oversight in the race to AGI
The ‘openai files’ push for oversight in the race to agi
📝 要約
“the openai files,” an archival project from the midas project and the tech oversight project, are a “collection of documented concerns with governanc...
Here’s your first look at the rebooted Digg
Here’s your first look at the rebooted digg
📝 要約
Digg launched its ios app to testers who are a part of its groundbreakers community of early adopters.
Seed to Series C: What VCs actually want from AI startups
Seed to series c: what vcs actually want from ai startups
📝 要約
Ai investments hit $110 billion in 2024, and the 資金調達 landscape in 2025 is more competitive than ever.
Google’s AI Mode can now have back-and-forth voice conversations
Google’s ai mode can now have back-and-forth voice conversations
📝 要約
Google is rolling out the ability for users to have a back-and-forth voice conversation with ai mode.
SportsVisio raises $3.2M for AI for sports athletes and fans
Sportsvisio raises $3.2m for ai for sports athletes and fans
📝 要約
Sportsvisio has raised $3.
Why AI hardware needs to be open
Why ai hardware needs to be open
📝 要約
When openai acquired io to create “the coolest piece of tech that the world will have ever seen,” it confirmed what industry experts have long been sa...
The Interpretable AI playbook: What Anthropic’s research means for your enterprise LLM strategy
The interpretable ai playbook: what anthropic’s 研究 means for your enterprise LLM strategy
📝 要約
Anthropic is developing “interpretable” ai, where models let us understand what they are thinking and arrive at a particular conclusion.
Sam Altman says Meta tried and failed to poach OpenAI’s talent with $100M offers
Sam altman says meta tried and failed to poach openai’s talent with $100m offers
📝 要約
Openai ceo sam altman said that meta tried to poach its employees with nine-figure offers, but failed to recruit openai's best people.
Google launches production-ready Gemini 2.5 AI models to challenge OpenAI’s enterprise dominance
Google ローンチ production-ready gemini 2.5 ai models to challenge openai’s enterprise dominance
📝 要約
Google ローンチ production-ready gemini 2.
OpenAI moves forward with GPT-4.5 deprecation in API, triggering developer anguish and confusion
Openai moves forward with GPT-4.5 deprecation in API, triggering developer anguish and confusion
📝 要約
Despite the strong reaction, openai had in fact already announced the plan to deprecate GPT-4.
OpenAI’s $200M DoD contract could squeeze frenemy Microsoft
Openai’s $200m dod contract could squeeze frenemy microsoft
📝 要約
The contract could put the モデル maker in competition with the openai services that microsoft wants to sell to the dod.
Google’s Gemini panicked when playing Pokémon
Google’s gemini panicked when playing pokémon
📝 要約
Ai companies are battling to dominate the industry, but sometimes they’re also battling in pokémon gyms.
Gemini 2.5: Updates to our family of thinking models
Gemini 2.5: updates to our family of thinking models
📝 要約
Explore the latest gemini 2.
We’re expanding our Gemini 2.5 family of models
We’re expanding our gemini 2.5 family of models
Meta buys stake in Scale AI, raising antitrust concerns
Meta buys stake in scale ai, raising antitrust concerns
Do reasoning AI models really ‘think’ or not? Apple research sparks lively debate, response
Do reasoning ai models really ‘think’ or not? apple 研究 sparks lively debate, response
📝 要約
Ultimately, the big takeaway for ml researchers is that before proclaiming an ai milestone—or obituary—make sure the test itself isn’t flawed
Beyond GPT architecture: Why Google’s Diffusion approach could reshape LLM deployment
Beyond GPT architecture: why google’s diffusion approach could reshape LLM deployment
📝 要約
Gemini diffusion is also useful for tasks such as refactoring code, adding new features to applications, or converting an existing codebase to a diffe...
Behind “ANCESTRA”: combining Veo with live-action filmmaking
Behind “ancestra”: combining veo with live-action filmmaking
📝 要約
We partnered with darren aronofsky, eliza mcnitt and a team of more than 200 people to make a film using veo and live-action filmmaking.
Meta’s new world model lets robots manipulate objects in environments they’ve never encountered before
Meta’s new world モデル lets robots manipulate objects in environments they’ve never encountered before
📝 要約
A robot powered by v-jepa 2 can be deployed in a new environment and successfully manipulate objects it has never encountered before.
How we're supporting better tropical cyclone prediction with AI
How we're supporting better tropical cyclone prediction with ai
📝 要約
We’re launching weather lab, featuring our experimental cyclone predictions, and we’re partnering with the u.
Advanced audio dialog and generation with Gemini 2.5
Advanced audio dialog and generation with gemini 2.5
Gemini 2.5: Our most intelligent models are getting even better
Gemini 2.5: our most intelligent models are getting even better
Our vision for building a universal AI assistant
Our vision for building a universal ai assistant
📝 要約
We’re extending gemini to become a world モデル that can make plans and imagine new experiences by simulating aspects of the world.
Advancing Gemini's security safeguards
Advancing gemini's セキュリティ safeguards
📝 要約
We’ve made gemini 2.
Announcing Gemma 3n preview: Powerful, efficient, mobile-first AI
Announcing gemma 3n preview: powerful, efficient, mobile-first ai
📝 要約
Gemma 3n is a cutting-edge open モデル designed for fast, multimodal ai on devices, featuring optimized performance, unique flexibility with a 2-in-1 m...
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
Alphaevolve: a gemini-powered coding agent for designing advanced algorithms
📝 要約
New ai agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated e...
Gemini 2.5 Pro Preview: even better coding performance
Gemini 2.5 pro preview: even better coding performance
📝 要約
We’ve seen developers doing amazing things with gemini 2.
Build rich, interactive web apps with an updated Gemini 2.5 Pro
Build rich, interactive web apps with an updated gemini 2.5 pro
📝 要約
Our updated version of gemini 2.
Introducing Gemini 2.5 Flash
Introducing gemini 2.5 flash
Generate videos in Gemini and Whisk with Veo 2
Generate videos in gemini and whisk with veo 2
📝 要約
Transform text-based prompts into high-resolution eight-second videos in gemini advanced and use whisk animate to turn images into eight-second animat...
DolphinGemma: How Google AI is helping decode dolphin communication
Dolphingemma: how google ai is helping decode dolphin communication
📝 要約
Dolphingemma, a 大規模言語モデル developed by google, is helping scientists study how dolphins communicate — and hopefully find out what they're s...
Gemini 2.5: Our most intelligent AI model
Gemini 2.5: our most intelligent ai モデル
Gemini Robotics brings AI into the physical world
Gemini ロボティクス brings ai into the physical world
📝 要約
Introducing gemini ロボティクス and gemini ロボティクス-er, ai models designed for robots to understand, act and react to the physical world.
Experiment with Gemini 2.0 Flash native image generation
Experiment with gemini 2.0 flash native image generation
📝 要約
Native image output is available in gemini 2.
Introducing Gemma 3
Introducing gemma 3
📝 要約
The most capable モデル you can run on a single gpu or tpu.
Start building with Gemini 2.0 Flash and Flash-Lite
Start building with gemini 2.0 flash and flash-lite
Gemini 2.0 is now available to everyone
Gemini 2.0 is now available to everyone
📝 要約
We’re announcing new updates to gemini 2.
FACTS Grounding: A new benchmark for evaluating the factuality of large language models
Facts grounding: a new benchmark for evaluating the factuality of large language models
📝 要約
Our comprehensive benchmark and online leaderboard offer a much-needed measure of how accurately llms ground their responses in provided source materi...
State-of-the-art video and image generation with Veo 2 and Imagen 3
State-of-the-art video and image generation with veo 2 and imagen 3
📝 要約
We’re rolling out a new, state-of-the-art video モデル, veo 2, and updates to imagen 3.
Introducing Gemini 2.0: our new AI model for the agentic era
Introducing gemini 2.0: our new ai モデル for the agentic era
📝 要約
Today, we’re announcing gemini 2.
Google DeepMind at NeurIPS 2024
Google deepmind at neurips 2024
📝 要約
Advancing adaptive ai agents, empowering 3d scene creation, and innovating LLM 学習 for a smarter, safer future
Genie 2: A large-scale foundation world model
Genie 2: a large-scale foundation world モデル
📝 要約
Generating unlimited diverse 学習 environments for future general agents
AlphaQubit tackles one of quantum computing’s biggest challenges
Alphaqubit tackles one of 量子コンピューティング’s biggest challenges
📝 要約
Our new ai system accurately identifies errors inside quantum computers, helping to make this new technology more reliable.
The AI for Science Forum: A new era of discovery
The ai for science forum: a new era of discovery
📝 要約
The ai science forum highlights ai's present and potential role in revolutionizing scientific discovery and solving global challenges, emphasizing col...
Updated production-ready Gemini models, reduced 1.5 Pro pricing, increased rate limits, and more
Updated production-ready gemini models, reduced 1.5 pro pricing, increased rate limits, and more
📝 要約
We’re releasing two updated production-ready gemini models
Our latest advances in robot dexterity
Our latest advances in robot dexterity
📝 要約
Two new ai systems, aloha unleashed and demostart, help robots learn to perform complex tasks that require dexterous movement
FermiNet: Quantum physics and chemistry from first principles
Ferminet: quantum physics and chemistry from first principles
📝 要約
Using ディープラーニング to solve fundamental problems in computational quantum chemistry and explore how matter interacts with light
AI achieves silver-medal standard solving International Mathematical Olympiad problems
Ai achieves silver-medal standard solving international mathematical olympiad problems
📝 要約
ブレークスルー models alphaproof and alphageometry 2 solve advanced reasoning problems in mathematics
Google DeepMind at ICML 2024
Google deepmind at icml 2024
📝 要約
Exploring agi, the challenges of scaling and the future of multimodal 生成AI
Gemini breaks new ground: a faster model, longer context and AI agents
Gemini breaks new ground: a faster モデル, longer context and ai agents
📝 要約
We’re introducing a series of updates across the gemini family of models, including the new 1.
Watermarking AI-generated text and video with SynthID
Watermarking ai-generated text and video with synthid
📝 要約
Announcing our novel watermarking method for ai-generated text and video, and how we’re bringing synthid to key google products
AlphaFold 3 predicts the structure and interactions of all of life’s molecules
Alphafold 3 predicts the structure and interactions of all of life’s molecules
📝 要約
Introducing a new ai モデル developed by google deepmind and isomorphic labs.
Google DeepMind at ICLR 2024
Google deepmind at iclr 2024
📝 要約
Developing next-gen ai agents, exploring new modalities, and pioneering foundational learning
Generative AI to quantify uncertainty in weather forecasting
生成AI to quantify uncertainty in weather forecasting
📝 要約
Posted by lizao (larry) li, software engineer, and rob carver, 研究 scientist, google 研究 accurate weather forecasts can have a direct impact...
AutoBNN: Probabilistic time series forecasting with compositional bayesian neural networks
Autobnn: probabilistic time series forecasting with compositional bayesian neural networks
📝 要約
Posted by urs köster, software engineer, google 研究 time series problems are ubiquitous, from forecasting weather and traffic patterns to underst...
Computer-aided diagnosis for lung cancer screening
Computer-aided 診断 for lung cancer screening
📝 要約
Posted by atilla kiraly, software engineer, and rory pilgrim, product manager, google 研究 lung cancer is the leading cause of cancer-related deat...
Using AI to expand global access to reliable flood forecasts
Using ai to expand global access to reliable flood forecasts
📝 要約
Posted by yossi matias, vp engineering & 研究, and grey nearing, 研究 scientist, google 研究 floods are the most common natural disaster, ...
ScreenAI: A visual language model for UI and visually-situated language understanding
Screenai: a visual language モデル for ui and visually-situated language understanding
📝 要約
Posted by srinivas sunkara and gilles baechler, software engineers, google 研究 screen user interfaces (uis) and infographics, such as charts, dia...
SCIN: A new resource for representative dermatology images
Scin: a new resource for representative dermatology images
📝 要約
Posted by pooja rao, 研究 scientist, google 研究 health datasets play a crucial role in 研究 and 医療 education, but it can be challeng...
MELON: Reconstructing 3D objects from images with unknown poses
Melon: reconstructing 3d objects from images with unknown poses
📝 要約
Posted by mark matthews, senior software engineer, and dmitry lagun, 研究 scientist, google 研究 a person's prior experience and understanding...
HEAL: A framework for health equity assessment of machine learning performance
Heal: a framework for health equity assessment of 機械学習 performance
📝 要約
Posted by mike schaekermann, 研究 scientist, google 研究, and ivor horn, chief health equity officer & director, google core health equity is ...
Cappy: Outperforming and boosting large multi-task language models with a small scorer
Cappy: outperforming and boosting large multi-task language models with a small scorer
📝 要約
Posted by yun zhu and lijuan liu, software engineers, google 研究 大規模言語モデル (LLM) advancements have led to a new paradigm that unifies ...
A generalist AI agent for 3D virtual environments
A generalist ai agent for 3d virtual environments
📝 要約
Introducing sima, a scalable instructable multiworld agent
Talk like a graph: Encoding graphs for large language models
Talk like a graph: encoding graphs for large language models
📝 要約
Posted by bahare fatemi and bryan perozzi, 研究 scientists, google 研究 imagine all the things around you — your friends, tools in your kitche...
Chain-of-table: Evolving tables in the reasoning chain for table understanding
Chain-of-table: evolving tables in the reasoning chain for table understanding
📝 要約
Posted by zilong wang, student researcher, and chen-yu lee, 研究 scientist, クラウド ai team people use tables every day to organize and interpret co...
Health-specific embedding tools for dermatology and pathology
Health-specific embedding tools for dermatology and pathology
📝 要約
Posted by dave steiner, clinical 研究 scientist, google health, and rory pilgrim, product manager, google 研究 there’s a worldwide shortage of...
Social learning: Collaborative learning with large language models
Social learning: collaborative learning with large language models
📝 要約
Posted by amirkeivan mohtashami, 研究 intern, and florian hartmann, software engineer, google 研究 large language models (llms) have significa...
Croissant: a metadata format for ML-ready datasets
Croissant: a metadata format for ml-ready datasets
📝 要約
Posted by omar benjelloun, software engineer, google 研究, and peter mattson, software engineer, google core ml and president, mlcommons associati...
Google at APS 2024
Google at aps 2024
📝 要約
Posted by kate weber and shannon leon, google 研究, quantum ai team today the 2024 march meeting of the american physical society (aps) kicks off ...
VideoPrism: A foundational visual encoder for video understanding
Videoprism: a foundational visual encoder for video understanding
📝 要約
Posted by long zhao, senior 研究 scientist, and ting liu, senior staff software engineer, google 研究 an astounding number of videos are avail...
Advances in private training for production on-device language models
Advances in private 学習 for production on-device language models
📝 要約
Posted by zheng xu, 研究 scientist, and yanxiang zhang, software engineer, google language models (lms) trained to predict the next word given inp...
Gemma: Introducing new state-of-the-art open models
Gemma: introducing new state-of-the-art open models
📝 要約
Gemma is built for responsible ai development from the same 研究 and technology used to create gemini models.
Our next-generation model: Gemini 1.5
Our next-generation モデル: gemini 1.5
📝 要約
The モデル delivers dramatically enhanced performance, with a ブレークスルー in long-context understanding across modalities.
Learning the importance of training data under concept drift
Learning the importance of 学習 data under concept drift
📝 要約
Posted by nishant jain, pre-doctoral researcher, and pradeep shenoy, 研究 scientist, google 研究 the constantly changing nature of the world a...
DP-Auditorium: A flexible library for auditing differential privacy
Dp-auditorium: a flexible library for auditing differential プライバシー
📝 要約
Posted by mónica ribero díaz, 研究 scientist, google 研究 differential プライバシー (dp) is a property of randomized mechanisms that limit the infl...
The next chapter of our Gemini era
The next chapter of our gemini era
📝 要約
We're bringing gemini to more google products
Graph neural networks in TensorFlow
Graph neural networks in tensorflow
📝 要約
Posted by dustin zelle, software engineer, google 研究, and arno eigenwillig, software engineer, coreml objects and their relationships are ubiqui...
A decoder-only foundation model for time-series forecasting
A decoder-only foundation モデル for time-series forecasting
📝 要約
Posted by rajat sen and yichen zhou, google 研究 time-series forecasting is ubiquitous in various domains, such as retail, finance, manufacturing,...
Intervening on early readouts for mitigating spurious features and simplicity bias
Intervening on early readouts for mitigating spurious features and simplicity バイアス
📝 要約
Posted by rishabh tiwari, pre-doctoral researcher, and pradeep shenoy, 研究 scientist, google 研究 機械学習 models in the real world a...
MobileDiffusion: Rapid text-to-image generation on-device
Mobilediffusion: rapid text-to-image generation on-device
📝 要約
Posted by yang zhao, senior software engineer, and tingbo hou, senior staff software engineer, core ml text-to-image diffusion models have shown excep...
Mixed-input matrix multiplication performance optimizations
Mixed-input matrix multiplication performance optimizations
📝 要約
Posted by manish gupta, staff software engineer, google 研究 ai-driven technologies are weaving themselves into the fabric of our daily routines, ...
Exphormer: Scaling transformers for graph-structured data
Exphormer: scaling transformers for graph-structured data
📝 要約
Posted by ameya velingker, 研究 scientist, google 研究, and balaji venkatachalam, software engineer, google graphs, in which objects and their...
Shaping the future of advanced robotics
Shaping the future of advanced ロボティクス
📝 要約
Introducing autort, sara-rt, and rt-trajectory
2023: A Year of Groundbreaking Advances in AI and Computing
2023: a year of groundbreaking advances in ai and computing
📝 要約
This has been a year of incredible progress in the field of 人工知能 (ai) 研究 and its practical applications.
FunSearch: Making new discoveries in mathematical sciences using Large Language Models
Funsearch: making new discoveries in mathematical sciences using large language models
📝 要約
In a 論文 published in nature, we introduce funsearch, a method for searching for “functions” written in computer code, and find new solutions in mat...
Google DeepMind at NeurIPS 2023
Google deepmind at neurips 2023
📝 要約
The neural information processing systems (neurips) is the largest 人工知能 (ai) conference in the world.
Introducing Gemini: our largest and most capable AI model
Introducing gemini: our largest and most capable ai モデル
📝 要約
Making ai more helpful for everyone
Scaling up learning across many different robot types
Scaling up learning across many different robot types
📝 要約
Robots are great specialists, but poor generalists.
RT-2: New model translates vision and language into action
Rt-2: new モデル translates vision and language into action
📝 要約
Robotic トランスフォーマー 2 (rt-2) is a novel vision-language-action (vla) モデル that learns from both web and ロボティクス data, and translates this knowledge ...
Google DeepMind’s latest research at ICML 2023
Google deepmind’s latest 研究 at icml 2023
📝 要約
Exploring ai safety, adaptability, and efficiency for the real world
Developing reliable AI tools for healthcare
Developing reliable ai tools for ヘルスケア
📝 要約
We’ve published our joint 論文 with google 研究 in nature medicine, which proposes codoc (complementarity-driven deferral-to-clinical workflow), ...
RoboCat: A self-improving robotic agent
Robocat: a self-improving robotic agent
📝 要約
Robots are quickly becoming part of our everyday lives, but they’re often only programmed to perform specific tasks well.
Google Cloud: Driving digital transformation
Google クラウド: driving digital transformation
📝 要約
Google クラウド empowers organizations to digitally transform themselves into smarter businesses.
DeepMind’s latest research at ICLR 2023
Deepmind’s latest 研究 at iclr 2023
📝 要約
Next week marks the start of the 11th international conference on learning representations (iclr), taking place 1-5 may in kigali, rwanda.
Announcing Google DeepMind
Announcing google deepmind
📝 要約
Deepmind and the brain team from google 研究 will join forces to accelerate progress towards a world in which ai helps solve the biggest challenge...
DeepMind’s latest research at NeurIPS 2022
Deepmind’s latest 研究 at neurips 2022
📝 要約
Neurips is the world’s largest conference in 人工知能 (ai) and 機械学習 (ml), and we’re proud to support the event as diamond s...
Gemini 2.5: Updates to our family of thinking models
Gemini 2.5: updates to our family of thinking models
📝 要約
Explore the latest gemini 2.
We’re expanding our Gemini 2.5 family of models
We’re expanding our gemini 2.5 family of models
Behind “ANCESTRA”: combining Veo with live-action filmmaking
Behind “ancestra”: combining veo with live-action filmmaking
📝 要約
We partnered with darren aronofsky, eliza mcnitt and a team of more than 200 people to make a film using veo and live-action filmmaking.
How we're supporting better tropical cyclone prediction with AI
How we're supporting better tropical cyclone prediction with ai
📝 要約
We’re launching weather lab, featuring our experimental cyclone predictions, and we’re partnering with the u.
Advanced audio dialog and generation with Gemini 2.5
Advanced audio dialog and generation with gemini 2.5
Gemini 2.5: Our most intelligent models are getting even better
Gemini 2.5: our most intelligent models are getting even better
Our vision for building a universal AI assistant
Our vision for building a universal ai assistant
📝 要約
We’re extending gemini to become a world モデル that can make plans and imagine new experiences by simulating aspects of the world.
Advancing Gemini's security safeguards
Advancing gemini's セキュリティ safeguards
📝 要約
We’ve made gemini 2.
Announcing Gemma 3n preview: Powerful, efficient, mobile-first AI
Announcing gemma 3n preview: powerful, efficient, mobile-first ai
📝 要約
Gemma 3n is a cutting-edge open モデル designed for fast, multimodal ai on devices, featuring optimized performance, unique flexibility with a 2-in-1 m...
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
Alphaevolve: a gemini-powered coding agent for designing advanced algorithms
📝 要約
New ai agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated e...
Gemini 2.5 Pro Preview: even better coding performance
Gemini 2.5 pro preview: even better coding performance
📝 要約
We’ve seen developers doing amazing things with gemini 2.
Build rich, interactive web apps with an updated Gemini 2.5 Pro
Build rich, interactive web apps with an updated gemini 2.5 pro
📝 要約
Our updated version of gemini 2.
Introducing Gemini 2.5 Flash
Introducing gemini 2.5 flash
Generate videos in Gemini and Whisk with Veo 2
Generate videos in gemini and whisk with veo 2
📝 要約
Transform text-based prompts into high-resolution eight-second videos in gemini advanced and use whisk animate to turn images into eight-second animat...
DolphinGemma: How Google AI is helping decode dolphin communication
Dolphingemma: how google ai is helping decode dolphin communication
📝 要約
Dolphingemma, a 大規模言語モデル developed by google, is helping scientists study how dolphins communicate — and hopefully find out what they're s...
Gemini 2.5: Our most intelligent AI model
Gemini 2.5: our most intelligent ai モデル
Gemini Robotics brings AI into the physical world
Gemini ロボティクス brings ai into the physical world
📝 要約
Introducing gemini ロボティクス and gemini ロボティクス-er, ai models designed for robots to understand, act and react to the physical world.
Experiment with Gemini 2.0 Flash native image generation
Experiment with gemini 2.0 flash native image generation
📝 要約
Native image output is available in gemini 2.
Introducing Gemma 3
Introducing gemma 3
📝 要約
The most capable モデル you can run on a single gpu or tpu.
Start building with Gemini 2.0 Flash and Flash-Lite
Start building with gemini 2.0 flash and flash-lite
Gemini 2.0 is now available to everyone
Gemini 2.0 is now available to everyone
📝 要約
We’re announcing new updates to gemini 2.
FACTS Grounding: A new benchmark for evaluating the factuality of large language models
Facts grounding: a new benchmark for evaluating the factuality of large language models
📝 要約
Our comprehensive benchmark and online leaderboard offer a much-needed measure of how accurately llms ground their responses in provided source materi...
State-of-the-art video and image generation with Veo 2 and Imagen 3
State-of-the-art video and image generation with veo 2 and imagen 3
📝 要約
We’re rolling out a new, state-of-the-art video モデル, veo 2, and updates to imagen 3.
Introducing Gemini 2.0: our new AI model for the agentic era
Introducing gemini 2.0: our new ai モデル for the agentic era
📝 要約
Today, we’re announcing gemini 2.
Google DeepMind at NeurIPS 2024
Google deepmind at neurips 2024
📝 要約
Advancing adaptive ai agents, empowering 3d scene creation, and innovating LLM 学習 for a smarter, safer future
Genie 2: A large-scale foundation world model
Genie 2: a large-scale foundation world モデル
📝 要約
Generating unlimited diverse 学習 environments for future general agents
AlphaQubit tackles one of quantum computing’s biggest challenges
Alphaqubit tackles one of 量子コンピューティング’s biggest challenges
📝 要約
Our new ai system accurately identifies errors inside quantum computers, helping to make this new technology more reliable.
The AI for Science Forum: A new era of discovery
The ai for science forum: a new era of discovery
📝 要約
The ai science forum highlights ai's present and potential role in revolutionizing scientific discovery and solving global challenges, emphasizing col...
Updated production-ready Gemini models, reduced 1.5 Pro pricing, increased rate limits, and more
Updated production-ready gemini models, reduced 1.5 pro pricing, increased rate limits, and more
📝 要約
We’re releasing two updated production-ready gemini models
Our latest advances in robot dexterity
Our latest advances in robot dexterity
📝 要約
Two new ai systems, aloha unleashed and demostart, help robots learn to perform complex tasks that require dexterous movement
FermiNet: Quantum physics and chemistry from first principles
Ferminet: quantum physics and chemistry from first principles
📝 要約
Using ディープラーニング to solve fundamental problems in computational quantum chemistry and explore how matter interacts with light
AI achieves silver-medal standard solving International Mathematical Olympiad problems
Ai achieves silver-medal standard solving international mathematical olympiad problems
📝 要約
ブレークスルー models alphaproof and alphageometry 2 solve advanced reasoning problems in mathematics
Google DeepMind at ICML 2024
Google deepmind at icml 2024
📝 要約
Exploring agi, the challenges of scaling and the future of multimodal 生成AI
Gemini breaks new ground: a faster model, longer context and AI agents
Gemini breaks new ground: a faster モデル, longer context and ai agents
📝 要約
We’re introducing a series of updates across the gemini family of models, including the new 1.
Watermarking AI-generated text and video with SynthID
Watermarking ai-generated text and video with synthid
📝 要約
Announcing our novel watermarking method for ai-generated text and video, and how we’re bringing synthid to key google products
AlphaFold 3 predicts the structure and interactions of all of life’s molecules
Alphafold 3 predicts the structure and interactions of all of life’s molecules
📝 要約
Introducing a new ai モデル developed by google deepmind and isomorphic labs.
Google DeepMind at ICLR 2024
Google deepmind at iclr 2024
📝 要約
Developing next-gen ai agents, exploring new modalities, and pioneering foundational learning
Generative AI to quantify uncertainty in weather forecasting
生成AI to quantify uncertainty in weather forecasting
📝 要約
Posted by lizao (larry) li, software engineer, and rob carver, 研究 scientist, google 研究 accurate weather forecasts can have a direct impact...
AutoBNN: Probabilistic time series forecasting with compositional bayesian neural networks
Autobnn: probabilistic time series forecasting with compositional bayesian neural networks
📝 要約
Posted by urs köster, software engineer, google 研究 time series problems are ubiquitous, from forecasting weather and traffic patterns to underst...
Computer-aided diagnosis for lung cancer screening
Computer-aided 診断 for lung cancer screening
📝 要約
Posted by atilla kiraly, software engineer, and rory pilgrim, product manager, google 研究 lung cancer is the leading cause of cancer-related deat...
Using AI to expand global access to reliable flood forecasts
Using ai to expand global access to reliable flood forecasts
📝 要約
Posted by yossi matias, vp engineering & 研究, and grey nearing, 研究 scientist, google 研究 floods are the most common natural disaster, ...
ScreenAI: A visual language model for UI and visually-situated language understanding
Screenai: a visual language モデル for ui and visually-situated language understanding
📝 要約
Posted by srinivas sunkara and gilles baechler, software engineers, google 研究 screen user interfaces (uis) and infographics, such as charts, dia...
SCIN: A new resource for representative dermatology images
Scin: a new resource for representative dermatology images
📝 要約
Posted by pooja rao, 研究 scientist, google 研究 health datasets play a crucial role in 研究 and 医療 education, but it can be challeng...
MELON: Reconstructing 3D objects from images with unknown poses
Melon: reconstructing 3d objects from images with unknown poses
📝 要約
Posted by mark matthews, senior software engineer, and dmitry lagun, 研究 scientist, google 研究 a person's prior experience and understanding...
HEAL: A framework for health equity assessment of machine learning performance
Heal: a framework for health equity assessment of 機械学習 performance
📝 要約
Posted by mike schaekermann, 研究 scientist, google 研究, and ivor horn, chief health equity officer & director, google core health equity is ...
Cappy: Outperforming and boosting large multi-task language models with a small scorer
Cappy: outperforming and boosting large multi-task language models with a small scorer
📝 要約
Posted by yun zhu and lijuan liu, software engineers, google 研究 大規模言語モデル (LLM) advancements have led to a new paradigm that unifies ...
A generalist AI agent for 3D virtual environments
A generalist ai agent for 3d virtual environments
📝 要約
Introducing sima, a scalable instructable multiworld agent
Talk like a graph: Encoding graphs for large language models
Talk like a graph: encoding graphs for large language models
📝 要約
Posted by bahare fatemi and bryan perozzi, 研究 scientists, google 研究 imagine all the things around you — your friends, tools in your kitche...
Chain-of-table: Evolving tables in the reasoning chain for table understanding
Chain-of-table: evolving tables in the reasoning chain for table understanding
📝 要約
Posted by zilong wang, student researcher, and chen-yu lee, 研究 scientist, クラウド ai team people use tables every day to organize and interpret co...
Health-specific embedding tools for dermatology and pathology
Health-specific embedding tools for dermatology and pathology
📝 要約
Posted by dave steiner, clinical 研究 scientist, google health, and rory pilgrim, product manager, google 研究 there’s a worldwide shortage of...
Social learning: Collaborative learning with large language models
Social learning: collaborative learning with large language models
📝 要約
Posted by amirkeivan mohtashami, 研究 intern, and florian hartmann, software engineer, google 研究 large language models (llms) have significa...
Croissant: a metadata format for ML-ready datasets
Croissant: a metadata format for ml-ready datasets
📝 要約
Posted by omar benjelloun, software engineer, google 研究, and peter mattson, software engineer, google core ml and president, mlcommons associati...
Google at APS 2024
Google at aps 2024
📝 要約
Posted by kate weber and shannon leon, google 研究, quantum ai team today the 2024 march meeting of the american physical society (aps) kicks off ...
VideoPrism: A foundational visual encoder for video understanding
Videoprism: a foundational visual encoder for video understanding
📝 要約
Posted by long zhao, senior 研究 scientist, and ting liu, senior staff software engineer, google 研究 an astounding number of videos are avail...
Advances in private training for production on-device language models
Advances in private 学習 for production on-device language models
📝 要約
Posted by zheng xu, 研究 scientist, and yanxiang zhang, software engineer, google language models (lms) trained to predict the next word given inp...
Gemma: Introducing new state-of-the-art open models
Gemma: introducing new state-of-the-art open models
📝 要約
Gemma is built for responsible ai development from the same 研究 and technology used to create gemini models.
Our next-generation model: Gemini 1.5
Our next-generation モデル: gemini 1.5
📝 要約
The モデル delivers dramatically enhanced performance, with a ブレークスルー in long-context understanding across modalities.
Learning the importance of training data under concept drift
Learning the importance of 学習 data under concept drift
📝 要約
Posted by nishant jain, pre-doctoral researcher, and pradeep shenoy, 研究 scientist, google 研究 the constantly changing nature of the world a...
DP-Auditorium: A flexible library for auditing differential privacy
Dp-auditorium: a flexible library for auditing differential プライバシー
📝 要約
Posted by mónica ribero díaz, 研究 scientist, google 研究 differential プライバシー (dp) is a property of randomized mechanisms that limit the infl...
The next chapter of our Gemini era
The next chapter of our gemini era
📝 要約
We're bringing gemini to more google products
Graph neural networks in TensorFlow
Graph neural networks in tensorflow
📝 要約
Posted by dustin zelle, software engineer, google 研究, and arno eigenwillig, software engineer, coreml objects and their relationships are ubiqui...
A decoder-only foundation model for time-series forecasting
A decoder-only foundation モデル for time-series forecasting
📝 要約
Posted by rajat sen and yichen zhou, google 研究 time-series forecasting is ubiquitous in various domains, such as retail, finance, manufacturing,...
Intervening on early readouts for mitigating spurious features and simplicity bias
Intervening on early readouts for mitigating spurious features and simplicity バイアス
📝 要約
Posted by rishabh tiwari, pre-doctoral researcher, and pradeep shenoy, 研究 scientist, google 研究 機械学習 models in the real world a...
MobileDiffusion: Rapid text-to-image generation on-device
Mobilediffusion: rapid text-to-image generation on-device
📝 要約
Posted by yang zhao, senior software engineer, and tingbo hou, senior staff software engineer, core ml text-to-image diffusion models have shown excep...
Mixed-input matrix multiplication performance optimizations
Mixed-input matrix multiplication performance optimizations
📝 要約
Posted by manish gupta, staff software engineer, google 研究 ai-driven technologies are weaving themselves into the fabric of our daily routines, ...
Exphormer: Scaling transformers for graph-structured data
Exphormer: scaling transformers for graph-structured data
📝 要約
Posted by ameya velingker, 研究 scientist, google 研究, and balaji venkatachalam, software engineer, google graphs, in which objects and their...
Shaping the future of advanced robotics
Shaping the future of advanced ロボティクス
📝 要約
Introducing autort, sara-rt, and rt-trajectory
2023: A Year of Groundbreaking Advances in AI and Computing
2023: a year of groundbreaking advances in ai and computing
📝 要約
This has been a year of incredible progress in the field of 人工知能 (ai) 研究 and its practical applications.
FunSearch: Making new discoveries in mathematical sciences using Large Language Models
Funsearch: making new discoveries in mathematical sciences using large language models
📝 要約
In a 論文 published in nature, we introduce funsearch, a method for searching for “functions” written in computer code, and find new solutions in mat...
Google DeepMind at NeurIPS 2023
Google deepmind at neurips 2023
📝 要約
The neural information processing systems (neurips) is the largest 人工知能 (ai) conference in the world.
Introducing Gemini: our largest and most capable AI model
Introducing gemini: our largest and most capable ai モデル
📝 要約
Making ai more helpful for everyone
Scaling up learning across many different robot types
Scaling up learning across many different robot types
📝 要約
Robots are great specialists, but poor generalists.
RT-2: New model translates vision and language into action
Rt-2: new モデル translates vision and language into action
📝 要約
Robotic トランスフォーマー 2 (rt-2) is a novel vision-language-action (vla) モデル that learns from both web and ロボティクス data, and translates this knowledge ...
Google DeepMind’s latest research at ICML 2023
Google deepmind’s latest 研究 at icml 2023
📝 要約
Exploring ai safety, adaptability, and efficiency for the real world
Developing reliable AI tools for healthcare
Developing reliable ai tools for ヘルスケア
📝 要約
We’ve published our joint 論文 with google 研究 in nature medicine, which proposes codoc (complementarity-driven deferral-to-clinical workflow), ...
RoboCat: A self-improving robotic agent
Robocat: a self-improving robotic agent
📝 要約
Robots are quickly becoming part of our everyday lives, but they’re often only programmed to perform specific tasks well.
Google Cloud: Driving digital transformation
Google クラウド: driving digital transformation
📝 要約
Google クラウド empowers organizations to digitally transform themselves into smarter businesses.
DeepMind’s latest research at ICLR 2023
Deepmind’s latest 研究 at iclr 2023
📝 要約
Next week marks the start of the 11th international conference on learning representations (iclr), taking place 1-5 may in kigali, rwanda.
Announcing Google DeepMind
Announcing google deepmind
📝 要約
Deepmind and the brain team from google 研究 will join forces to accelerate progress towards a world in which ai helps solve the biggest challenge...
DeepMind’s latest research at NeurIPS 2022
Deepmind’s latest 研究 at neurips 2022
📝 要約
Neurips is the world’s largest conference in 人工知能 (ai) and 機械学習 (ml), and we’re proud to support the event as diamond s...
A timeline of the US semiconductor market in 2025
A timeline of the us semiconductor market in 2025
📝 要約
It’s already been a tumultuous year for the u.
Here are the 24 US AI startups that have raised $100M or more in 2025
Here are the 24 us ai startups that have raised $100m or more in 2025
📝 要約
Last year was monumental for the ai industry in the u.
‘Kid-pilled’ Sam Altman ‘constantly’ asked ChatGPT questions about his newborn
‘kid-pilled’ sam altman ‘constantly’ asked chatgpt questions about his newborn
📝 要約
Sam altman, who is both the father of a 3-month-old and ceo of openai, hopped on openai's new podcast today to talk about how his company is impacting...
Six-month-old, solo-owned vibe coder Base44 sells to Wix for $80M cash
Six-month-old, solo-owned vibe coder base44 sells to wix for $80m cash
📝 要約
In base44's six months as a stand-alone company, it reportedly grew to 250,000 users and was making nearly $200,000 in monthly profits.
Multiplier, founded by ex-Stripe exec, nabs $27.5M to fuel AI-powered accounting roll-ups
Multiplier, founded by ex-stripe exec, nabs $27.5m to fuel ai-powered accounting roll-ups
📝 要約
Multiplier raised series a and seed 資金調達 from lightspeed and ribbit capital.
Midjourney launches its first AI video generation model, V1
Midjourney ローンチ its first ai video generation モデル, v1
📝 要約
Midjourney launched its much anticipated ai video generation モデル, v1, on wednesday and laid out a roadmap for its future releases.
OpenAI drops Scale AI as a data provider following Meta deal
Openai drops scale ai as a data provider following meta deal
📝 要約
Openai says it's cutting ties with scale ai following its deal with meta, raising questions about the data provider's business.
xAI is facing a lawsuit for operating over 400 MW of gas turbines without permits
Xai is facing a lawsuit for operating over 400 mw of gas turbines without permits
📝 要約
The colossus data center was allegedly operating as many as 35 gas turbines without pollution controls in a region that already had poor air quality.
OpenAI found features in AI models that correspond to different ‘personas’
Openai found features in ai models that correspond to different ‘personas’
📝 要約
By looking at an ai モデル's internal representations — the numbers that dictate how an ai モデル responds, which often seem completely incoherent to hu...
The ‘OpenAI Files’ push for oversight in the race to AGI
The ‘openai files’ push for oversight in the race to agi
📝 要約
“the openai files,” an archival project from the midas project and the tech oversight project, are a “collection of documented concerns with governanc...
Here’s your first look at the rebooted Digg
Here’s your first look at the rebooted digg
📝 要約
Digg launched its ios app to testers who are a part of its groundbreakers community of early adopters.
Seed to Series C: What VCs actually want from AI startups
Seed to series c: what vcs actually want from ai startups
📝 要約
Ai investments hit $110 billion in 2024, and the 資金調達 landscape in 2025 is more competitive than ever.
Google’s AI Mode can now have back-and-forth voice conversations
Google’s ai mode can now have back-and-forth voice conversations
📝 要約
Google is rolling out the ability for users to have a back-and-forth voice conversation with ai mode.
Sam Altman says Meta tried and failed to poach OpenAI’s talent with $100M offers
Sam altman says meta tried and failed to poach openai’s talent with $100m offers
📝 要約
Openai ceo sam altman said that meta tried to poach its employees with nine-figure offers, but failed to recruit openai's best people.
OpenAI’s $200M DoD contract could squeeze frenemy Microsoft
Openai’s $200m dod contract could squeeze frenemy microsoft
📝 要約
The contract could put the モデル maker in competition with the openai services that microsoft wants to sell to the dod.
Google’s Gemini panicked when playing Pokémon
Google’s gemini panicked when playing pokémon
📝 要約
Ai companies are battling to dominate the industry, but sometimes they’re also battling in pokémon gyms.
Google’s Gemini transparency cut leaves enterprise developers ‘debugging blind’
Google’s gemini transparency cut leaves enterprise developers ‘debugging blind’
📝 要約
Why is google hiding gemini's reasoning traces?
Unlock the other 99% of your data – now ready for AI
Unlock the other 99% of your data – now ready for ai
📝 要約
For decades, companies of all sizes have recognized that the data available to them holds significant value, for improving user and customer experienc...
It’s pretty easy to get DeepSeek to talk dirty
It’s pretty easy to get deepseek to talk dirty
📝 要約
Ai companions like replika are designed to engage in intimate exchanges, but people use general-purpose chatbots for sex talk too, despite their stric...
The OpenAI Files: Ex-staff claim profit greed betraying AI safety
The openai files: ex-staff claim profit greed betraying ai safety
📝 要約
‘the openai files’ report, assembling voices of concerned ex-staff, claims the world’s most prominent ai lab is betraying safety for profit.
GenLayer launches a new method to incentivize people to market your brand using AI and blockchain
Genlayer ローンチ a new method to incentivize people to market your brand using ai and blockchain
📝 要約
With applications like rally already live in beta, genlayer presents a new category of intelligent blockchain infrastructure.
Apple hints at AI integration in chip design process
Apple hints at ai integration in chip design process
📝 要約
Apple is beginning to use generative 人工知能 to help design the chips that power its devices.
Announcing our 2025 VB Transform Innovation Showcase finalists
Announcing our 2025 vb transform innovation showcase finalists
📝 要約
Seven companies will be sharing their latest ai innovations from the main stage at vb transform in sf on june 25.
OpenAI open sourced a new Customer Service Agent framework — learn more about its growing enterprise strategy
Openai open sourced a new customer service agent framework — learn more about its growing enterprise strategy
📝 要約
By offering transparent tooling and clear implementation examples, openai is pushing agentic systems out of the lab and into everyday use.
Announcing the 2025 finalists for VentureBeat Women in AI Awards
Announcing the 2025 finalists for venturebeat women in ai awards
📝 要約
Announcing the finalists for the 2025 women in ai awards.
OpenAI can rehabilitate AI models that develop a “bad boy persona”
Openai can rehabilitate ai models that develop a “bad boy persona”
📝 要約
A new 論文 from openai has shown why a little bit of bad 学習 can make ai models go rogue—but also demonstrates that this problem is generally pr...
SportsVisio raises $3.2M for AI for sports athletes and fans
Sportsvisio raises $3.2m for ai for sports athletes and fans
📝 要約
Sportsvisio has raised $3.
Why AI hardware needs to be open
Why ai hardware needs to be open
📝 要約
When openai acquired io to create “the coolest piece of tech that the world will have ever seen,” it confirmed what industry experts have long been sa...
The Interpretable AI playbook: What Anthropic’s research means for your enterprise LLM strategy
The interpretable ai playbook: what anthropic’s 研究 means for your enterprise LLM strategy
📝 要約
Anthropic is developing “interpretable” ai, where models let us understand what they are thinking and arrive at a particular conclusion.
Google launches production-ready Gemini 2.5 AI models to challenge OpenAI’s enterprise dominance
Google ローンチ production-ready gemini 2.5 ai models to challenge openai’s enterprise dominance
📝 要約
Google ローンチ production-ready gemini 2.
OpenAI moves forward with GPT-4.5 deprecation in API, triggering developer anguish and confusion
Openai moves forward with GPT-4.5 deprecation in API, triggering developer anguish and confusion
📝 要約
Despite the strong reaction, openai had in fact already announced the plan to deprecate GPT-4.
Meta buys stake in Scale AI, raising antitrust concerns
Meta buys stake in scale ai, raising antitrust concerns
Do reasoning AI models really ‘think’ or not? Apple research sparks lively debate, response
Do reasoning ai models really ‘think’ or not? apple 研究 sparks lively debate, response
📝 要約
Ultimately, the big takeaway for ml researchers is that before proclaiming an ai milestone—or obituary—make sure the test itself isn’t flawed
Beyond GPT architecture: Why Google’s Diffusion approach could reshape LLM deployment
Beyond GPT architecture: why google’s diffusion approach could reshape LLM deployment
📝 要約
Gemini diffusion is also useful for tasks such as refactoring code, adding new features to applications, or converting an existing codebase to a diffe...
Meta’s new world model lets robots manipulate objects in environments they’ve never encountered before
Meta’s new world モデル lets robots manipulate objects in environments they’ve never encountered before
📝 要約
A robot powered by v-jepa 2 can be deployed in a new environment and successfully manipulate objects it has never encountered before.