AI 분야의 주요 뉴스와 변화를 타임라인으로 확인하세요.
세상이 어떻게 바뀌고 있는지 한눈에 파악할 수 있습니다.
The suspect in the mass shooting at Tumbler Ridge, British Columbia, Jesse Van Rootselaar, was raising alarms among employees at OpenAI months before the shooting took place. This past June, Jesse had conversations with ChatGPT involving descriptions of gun violence that triggered the chatbot's auto
The Verge AISarvam's Indus chat app is currently available in beta.
TechCrunch AIJesse Van Rootselaar's descriptions of gun violence were flagged by tools that monitor ChatGPT for misuse.
TechCrunch AIAs generative AI evolves, a Google VP warns that LLM wrappers and AI aggregators face mounting pressure, with shrinking margins and limited differentiation threatening their long-term viability.
TechCrunch AIIs Microsoft's gaming division doubling down on AI?
TechCrunch AIHacker News에서 146점을 받은 글. 105개의 댓글.
Hacker NewsHacker News에서 161점을 받은 글. 87개의 댓글.
Hacker NewsHacker News에서 233점을 받은 글. 56개의 댓글.
Hacker NewsHacker News에서 285점을 받은 글. 726개의 댓글.
Hacker NewsOpenAI's first hardware release will be a smart speaker with a camera that will probably cost between $200 and $300, according to The Information. The device will be able to recognize things like "items on a nearby table or conversations people are having in the vicinity," The Information says, and
The Verge AIAmazon Web Services suffered a 13-hour outage to one system in December as a result of its AI coding assistant Kiro's actions, according to the Financial Times. Numerous unnamed Amazon employees told the FT that AI agent Kiro was responsible for the December incident affecting an AWS service in part
The Verge AIThe Trump administration just tossed out Biden-era restrictions on mercury and other toxic pollutants from power plants. It's repealing Mercury and Air Toxics Standards (MATS) just as electricity demand in the US ticks up with the buildout of new AI data centers. Those standards are particularly imp
The Verge AINvidia is working with investors, nonprofits, and venture firms to build earlier ties with India's fast-growing AI founder ecosystem.
TechCrunch AIGemini 3.1 Pro promises a Google LLM capable of handling more complex forms of work.
TechCrunch AIThe company said on Friday that users between 18 and 24 years of age account for nearly 50% of all messages sent by Indians to ChatGPT, and users under 30 account for 80% of usage in the country.
TechCrunch AIPeak XV says most of its new capital will target India as the firm prioritizes AI, fintech and cross-border bets while navigating recent partner departures.
TechCrunch AIAI expands access to filmmaking for resource-constrained creators. But as efficiency becomes the industry’s north star, creativity risks being overwhelmed by a deluge of low-effort, AI-generated content.
TechCrunch AIAddictive, AI-enabled tablets are taking over, and also, Woody is balding in the new Toy Story movie, out June 19.
TechCrunch AIA new report from Business Insider reveals that high-level engineers at xAI were pulled off other projects to make sure Grok could answer detailed questions about the video game Baldur's Gate.
TechCrunch AIThe startup, founded by accountants who worked at Flexport, Miro, Hopin and Thrive Global, automates the difficulties of prepping financial statements.
TechCrunch AIDueling pro-AI PACs have centered around backing or targeting one New York congressional bid: Alex Bores, whose RAISE Act requires AI developers to disclose safety protocols and report serious system misuse.
TechCrunch AIThe creator economy is evolving fast, and ad revenue alone isn’t cutting it anymore. YouTubers are launching product lines, acquiring startups, and building actual business empires. In fact, MrBeast’s company bought fintech startup Step,
TechCrunch AIHacker News에서 225점을 받은 글. 227개의 댓글.
Hacker NewsHacker News에서 815점을 받은 글. 218개의 댓글.
Hacker NewsThank you for joining us on the “Enterprise AI hub.” In this episode of the Infosys Knowledge Institute Podcast, Dylan Cosper speaks with Prasad Banala, director of software engineering at a large US-based retail organization, about operationalizing agentic AI across the software develop
MIT Tech ReviewThis is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. How uncrewed narco subs could transform the Colombian drug trade For decades, handmade narco subs have been some of the cocaine trade’s most elusive a
MIT Tech ReviewAI-enabled deception now permeates our online lives. There are the high-profile cases you may easily spot, like when White House officials recently shared a manipulated image of a protester in Minnesota and then mocked those asking about it. Other times, it slips quietly into social media feeds and
MIT Tech ReviewToday on Decoder we’re going to talk about the war for AI talent. Right now, the hottest job market on the planet is for AI researchers. The vast majority of these people are concentrated into a small number of hugely valuable, extremely fast-growing companies in the San Francisco Bay Area. No
The Verge AIA hacker tricked a popular AI coding tool into installing OpenClaw - the viral, open-source AI agent OpenClaw that "actually does things" - absolutely everywhere. Funny as a stunt, but a sign of what to come as more and more people let autonomous software use their computers on their behalf. The hac
The Verge AIEach episode of HBO's The Pitt features some degree of medical trauma that almost makes the hospital drama feel like a horror series. Some patients are dealing with gnarly lacerations while others are fighting off vicious blood infections that could rob them of their limbs, and the chaos of working
The Verge AIHacker News에서 639점을 받은 글. 209개의 댓글.
Hacker NewsGoogle has given Gemini the ability to spit out AI-generated music, courtesy of DeepMind's latest audio model. Beta access to Lyria 3 is rolling out in the Gemini app, enabling users to generate 30-second tracks based on text, images, and videos, without having to leave the chatbot window. The new m
The Verge AIHacker News에서 366점을 받은 글. 235개의 댓글.
Hacker NewsAnthropic in talks to raise $10 billion at $350 billion valuation led by GIC and Coatue Management. NVIDIA and Microsoft plan to invest up to $15 billion in exchange for $30 billion in compute capacity.
Web ResearchOpenAI laying groundwork for IPO that could value it at $1 trillion, with filing potentially coming in H2 2026. Company expects $20B in annualized revenue this year, up from $3.7B last year.
Web ResearchChatGPT is using Grokipedia as a source, and it's not the only AI tool to do so. Citations to Elon Musk's AI-generated encyclopedia are starting to appear in answers from Google's AI Overviews, AI Mode, and Gemini, too. Data suggests that's on the rise, heightening concerns about accuracy and misinf
The Verge AINvidia's CEO Jensen Huang denied reports that he was unhappy with OpenAI and said his company still planned to make a "huge" investment in the ChatGPT firm. NVIDIA announced in September that it would invest up to $100 billion in OpenAI, but recently, there have been suggestions that the deal might
The Verge AIIf co-founder Lachy Groom has any doubts, he doesn’t show it. He’s working with people who've been working on this problem for decades and who believe the timing is finally right, which is all he needs to know.
TechCrunch AINvidia CEO Jensen Huang said that a recent report of friction between his company and OpenAI was “nonsense.”
TechCrunch AIHacker News에서 98점을 받은 글. 81개의 댓글.
Hacker NewsHacker News에서 127점을 받은 글. 52개의 댓글.
Hacker NewsHacker News에서 347점을 받은 글. 275개의 댓글.
Hacker NewsHacker News에서 444점을 받은 글. 135개의 댓글.
Hacker NewsThe $100 billion megadeal between OpenAI and NVIDIA has stalled amid negotiations, though NVIDIA CEO Jensen Huang denied reports of friction and called them 'nonsense'.
Hacker NewsFor the last couple of years, I’ve been following the progress of a group of individuals who believe death is humanity’s “core problem.” Put simply, they say death is wrong—for everyone. They’ve even said it’s morally wrong. They established what they consider a new philosophy, and they called it Vi
MIT Tech ReviewThis is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. DHS is using Google and Adobe AI to make videos The news: The US Department of Homeland Security is using AI video generators from Google and Adobe to
MIT Tech ReviewCivitai—an online marketplace for buying and selling AI-generated content, backed by the venture capital firm Andreessen Horowitz—is letting users buy custom instruction files for generating celebrity deepfakes. Some of these files were specifically designed to make pornographic images banned by the
MIT Tech ReviewYes, you read that right. "Moltbook" is a social network of sorts for AI agents, particularly ones offered by OpenClaw (a viral AI assistant project that was formerly known as Moltbot, and before that, known as Clawdbot - until a legal dispute with Anthropic). Moltbook, which is set up similarly to
The Verge AIThe stock prices of some major video game companies, including Take-Two Interactive, Roblox, and Unity, had notable declines on Friday, just a day after Google announced its Project Genie tool that lets users prompt AI to generate interactive experiences, Reuters reports. Take-Two's stock price clos
The Verge AIThe company says you can use plug-ins to "tell Claude how you like work done, which tools and data to pull from, how to handle critical workflows, and what slash commands to expose so your team gets more consistent outcomes."
TechCrunch AIThe viral personal AI assistant formerly known as Clawdbot has a new shell — again. After briefly rebranding as Moltbot, it has now picked OpenClaw as its new name.
TechCrunch AIHacker News에서 99점을 받은 글. 33개의 댓글.
Hacker NewsHacker News에서 143점을 받은 글. 73개의 댓글.
Hacker NewsHacker News에서 175점을 받은 글. 60개의 댓글.
Hacker NewsOpenAI announced retirement of GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT on February 13, 2026 as 'the vast majority of usage has shifted to GPT-5.2'.
Web ResearchThis is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Meet the Vitalists: the hardcore longevity enthusiasts who believe death is “wrong” Last April, an excited crowd gathered at a compound in Berkeley, C
MIT Tech ReviewThe US Department of Homeland Security is using AI video generators from Google and Adobe to make and edit content shared with the public, a new document reveals. It comes as immigration agencies have flooded social media with content to support President Trump’s mass deportation agenda—some o
MIT Tech ReviewEveryone is panicking because AI is very bad; everyone is panicking because AI is very good. It’s just that you never know which one you’re going to get. Grok is a pornography machine. Claude Code can do anything from building websites to reading your MRI. So of course Gen Z is spooked by what this&
MIT Tech ReviewThis week, a new generative AI tool from Google let me create bad knockoffs of 3D Nintendo worlds. Check out my version of something like Super Mario 64: I didn't like Metroid Prime 4: Beyond, but it's better than my version of a Metroid Prime experience: Or how about my take on The Legend of [̷
The Verge AIWhile generative AI is being adopted at various levels of game development, a new survey suggests that developers increasingly think the technology is bad for the industry. According to the most recent survey from the Game Developers Conference, 52 percent of respondents said that gen AI is having a
The Verge AIApple's biggest acquisition ever is still its $3 billion Beats buy in 2014, but now the second biggest deal is bringing in Q.ai, a four-year-old AI audio startup. Apple did not disclose the terms, but Financial Times reports that Apple is spending $2 billion on the company. It also mentioned Q.ai pa
The Verge AIElon Musk-led SpaceX may merge with another company helmed by Elon Musk. Reuters initially reported Thursday that SpaceX and xAI are "in discussions to merge" ahead of SpaceX's IPO that's planned for later this year. The merger would help with SpaceX's plans to launch data centers into space, accord
The Verge AIWhen you're walking around and navigating with Maps, you can now ask Gemini questions like "What neighborhood am I in?" or "What are top-rated restaurants nearby?"
TechCrunch AIA new AI lab called Flapping Airplanes launched yesterday, and a Sequoia partner has an interesting take on why they stand out.
TechCrunch AIWith the highest global usage of Gemini for learning, India is emerging as a key market shaping Google's approach to AI in education.
TechCrunch AIOriginally, these music publishers had filed a lawsuit against Anthropic over its use of about 500 copyrighted works.
TechCrunch AINew data suggests the app is seeing declines in both app downloads and consumer spending, as the early hype about the AI video social network wears off.
TechCrunch AIStarting Thursday, Google AI Ultra subscribers in the U.S. can play around with Project Genie, which is powered by a combination of Google’s latest world model Genie 3, its image generation model Nano Banana Pro, and Gemini.
TechCrunch AIQ.ai is an Israeli startup specializing in imaging and machine learning, particularly technologies that enable devices to interpret whispered speech and enhance audio in noisy environments.
TechCrunch AIWith Microsoft spending many billions on data centers, and rumors that no one is using its AI, CEO Satya Nadella shared some usage numbers.
TechCrunch AIIf a deal materializes, it would mean Amazon is backing competing startups in the race for AI supremacy.
TechCrunch AIThis merger would bring the Grok chatbot, Starlink satellites, and SpaceX rockets together under one corporation.
TechCrunch AIOne intrepid Morgan Stanley analyst dared to ask how Apple will make money on its AI investments. The answer will not shock you.
TechCrunch AIHacker News에서 104점을 받은 글. 14개의 댓글.
Hacker NewsHacker News에서 168점을 받은 글. 167개의 댓글.
Hacker NewsThe ability to remember you and your preferences is rapidly becoming a big selling point for AI chatbots and agents.  Earlier this month, Google announced Personal Intelligence, a new way for people to interact with the company’s Gemini chatbot that draws on their Gmail, photos, search, and You
MIT Tech ReviewAI is driving unprecedented investment for massive data centers and an energy supply that can support its huge computational appetite. One potential source of electricity for these facilities is next-generation nuclear power plants, which could be cheaper to construct and safer to operate than their
MIT Tech ReviewElon Musk's AI company xAI disclosed earlier this month it had raised $20 billion.
TechCrunch AIMark Zuckerberg says 2026 will be "a big year for delivering personal super intelligence."
TechCrunch AIDeepSeek's next flagship V4 model confirmed for mid-February launch, featuring Engram conditional memory for 1M+ token context and coding-focused capabilities outperforming Claude and GPT-4o in internal tests.
Web ResearchEU announced over 307 million euros investment into AI and related technologies to strengthen European AI capabilities against US and China.
Deep ResearchMeta pays $14.3 billion for a 49% stake in Scale AI, marking one of the largest AI acquisitions of 2025-2026.
Web ResearchYann LeCun left Meta after 12 years to launch AMI Labs (Advanced Machine Intelligence), raising €500M at €3B valuation to develop JEPA-based world models as alternative to LLMs.
Web ResearchAdobe expanded Firefly Foundry with partnerships from CAA, WME, and major VFX studios for IP-safe custom generative AI models in filmmaking.
Web ResearchDeepSeek revealed MODEL1 architecture in GitHub FlashMLA update with 28 references across 114 files, showing restructured KV cache and FP8 decoding for V4 foundation.
Web ResearchOpenAI closes a SoftBank-led $40 billion round at a $300 billion post-money valuation, the largest AI funding round in history.
Web ResearchAnthropic donated Model Context Protocol (MCP), 'USB-C for AI', to Linux Foundation's Agentic AI Foundation. OpenAI and Microsoft embraced the standard.
Deep ResearchCalifornia AG Rob Bonta issued cease-and-desist to xAI, ordering Grok to stop creating non-consensual explicit deepfakes including nude images of women and children.
Web ResearchOpenAI launched ChatGPT Go, a new lightweight mobile-first ChatGPT experience optimized for quick interactions.
Web ResearchLloyds Banking Group launched AI Academy to achieve 100% AI literacy across all 60,000+ employees, one of the largest corporate AI training initiatives in finance.
Web ResearchAnthropic appointed Irina Ghose as Managing Director of India ahead of a Bengaluru office opening, expanding its global presence.
Web ResearchNVIDIA debuts Nemotron 3 family of open models, with Super and Ultra variants expected to be available in the first half of 2026.
Web ResearchAnthropic released new economic research building blocks to help researchers and policymakers understand AI's economic impacts.
Web ResearchTesco entered 3-year strategic partnership with Mistral AI for AI across customer-facing and internal operations, including a joint 'AI lab' for retail innovation.
Web ResearchAnthropic launched Claude Cowork with advanced file-editing for mainstream users, building on Claude Code architecture without technical barriers.
Deep ResearchApple officially partnered with Google to use 1.2 trillion parameter Gemini AI model as the foundation for the next-gen Siri, targeting March 2026 release with iOS 26.4.
Web ResearchNew research provides mathematical proof that large language models have fundamental limitations, claiming they are 'incapable of carrying out computational and agentic tasks beyond a certain complexity.'
Web ResearchxAI completed $20 billion Series E round exceeding initial $15B target, backed by NVIDIA, Cisco, Qatar Investment Authority. Funds for Colossus supercomputer expansion and Grok 5 development.
Web ResearchMeta announced contracts with Vistra, TerraPower, Oklo for 6.6 gigawatts of nuclear energy to power Prometheus AI Supercluster in New Albany, Ohio.
Deep ResearchLMArena, AI model evaluation platform, raised $150 million led by Felicis Ventures and UC Investments at $1.7 billion valuation.
Deep ResearchAgentic AI market projected to grow from $5.2B in 2024 to $200B by 2034, with focus shifting to smaller task-specific models as MCP becomes industry standard.
Web ResearchOpenAI announced OpenAI for Healthcare, products to help healthcare organizations scale care, reduce admin work, and power custom clinical solutions.
Deep ResearchAnthropic donated Model Context Protocol (MCP) to new Agentic AI Foundation (AAIF) spun out by Linux Foundation. OpenAI contributed AGENTS.md to the foundation.
Web ResearchLG unveiled CLOiD AI humanoid robot at CES 2026 for 'Zero Labor Home' vision, featuring dual 7-DOF arms trained via NVIDIA Isaac in digital twin environments for household tasks.
Web ResearchOpenAI introduced ChatGPT Health allowing secure connection to medical records and wellness apps including Apple Health, MyFitnessPal, Peloton. 230M+ users ask health questions weekly.
Deep ResearchFei-Fei Li's World Labs launched Marble, first commercial world model for spatial AI understanding in physical environments.
Web ResearchMeta CTO Andrew Bosworth announces at WEF Davos that Meta Superintelligence Labs team has delivered its first high-profile AI models internally, showing significant promise.
Web ResearchElon Musk's xAI completed $20 billion Series E with Nvidia, Cisco, Fidelity, Qatar Investment Authority to accelerate Grok 5 development.
Deep ResearchLMArena secured $150 million Series A funding led by Felicis and UC Investments, with backing from a16z and Kleiner Perkins, bringing valuation to $1.7 billion.
Web ResearchIndustry experts predict 2026 as year AI moves from brute-force scaling to practical deployment with smaller task-focused SLMs offering 10-30x efficiency gains.
Web ResearchNVIDIA unveiled Rubin platform at CES 2026 with six new chips (Vera CPU, Rubin GPU, NVLink 6, ConnectX-9, BlueField-4, Spectrum-6) delivering 10x lower cost per token and 4x fewer GPUs for MoE training.
Web ResearchMajor AI labs including Anthropic, OpenAI, xAI, Google, Meta, AWS confirmed adoption of NVIDIA Rubin platform. Dario Amodei: 'Efficiency gains enable longer memory, better reasoning.'
Web ResearchDeepSeek expanded R1 whitepaper from 22 to 86 pages with complete training pipeline, 20+ benchmark data, and technical appendix ahead of V4 launch.
Web ResearchBoston Dynamics humanoid robot Atlas began first field test at Hyundai's plant near Savannah, Georgia, autonomously performing roof rack sorting tasks.
Deep ResearchTDTA took effect requiring GenAI developers to publicly disclose training dataset information, with xAI already challenging the law.
Web ResearchCiti built internal AI workforce of 4,000 employees through volunteer programs, with 70%+ using approved AI tools daily for summarization, drafting, and coding.
Web ResearchAI pioneer Yann LeCun announces departure from Meta to start his own world model lab, reportedly seeking a $5 billion valuation.
Web ResearchAnthropic closes its second major round of 2025, bringing total funding to $16.5 billion across two rounds, with valuation reaching $183 billion.
Web ResearchMistral AI announced Mistral 3 using DeepSeek V3 architecture, demonstrating growing influence of Chinese open-source AI on Western development.
Deep ResearchAlphabet completes acquisition of Intersect for $4.75 billion as part of major AI infrastructure expansion.
Web ResearchSoftBank, OpenAI, and Oracle announce Stargate joint venture with up to $500 billion to build AI infrastructure in the U.S.
Web ResearchOpenAI launched GPT-5.2 after GPT-5.1 setback, ranking #1 on Artificial Analysis Intelligence Index v4.0 with extended reasoning capabilities.
Web ResearchOpenAI introduced GPT-5.2-Codex optimized for software engineering tasks amid intense competition from Gemini 3 and 'Code Red' scramble.
Deep ResearchGoogle released Gemini 3 Flash as fast, cost-effective version of Gemini 3, making it the default model in Gemini app.
Deep ResearchOpenAI opened app submissions for ChatGPT, enabling third-party services to run natively. Early pilot apps include Canva, Spotify, Zillow, Booking.com.
Deep ResearchEuropean Commission published first draft of Code of Practice for marking and labeling AI-generated content, expected finalized by June 2026.
Deep ResearchGoogle released Gemini 3 Flash, combining reasoning ability with speed and economy. Gemini 3 Pro outperformed major AI models in 19 of 20 benchmarks, reportedly triggering OpenAI 'code red'.
Web ResearchAnthropic donates Model Context Protocol (MCP), the 'USB-C for AI' that lets AI agents talk to external tools, to the Linux Foundation's new Agentic AI Foundation. OpenAI and Microsoft embrace MCP.
Web ResearchOpenAI releases GPT-5.2 with 400,000 token context window, completing the wave of frontier model releases in late 2025.
Web ResearchDisney announced $1 billion investment in OpenAI to allow users to generate 200+ copyrighted characters including Disney, Pixar, Marvel, Star Wars on Sora 2.
Deep ResearchPresident Trump signed executive order aiming to limit state AI laws, setting up conflicts between federal and state governments over AI governance.
Deep ResearchNVIDIA announced plans to release a 500 billion parameter model, marking its entry into frontier AI model development.
Web ResearchAnthropic announced Claude Code integration with Slack, enabling developers to access AI coding capabilities directly in workplace communication.
Deep ResearchOpenAI declares 'Code Red' in response to Gemini 3's competitive capabilities, delaying work on new initiatives to focus on competition with Google.
Web ResearchElon Musk's xAI raises $10 billion at a $200 billion valuation, led by Middle Eastern investors and sovereign wealth funds, putting it in direct competition with OpenAI and Anthropic.
Web ResearchRunway released Gen-4.5 video generation model, taking #1 on Video Arena leaderboard, surpassing Google Veo 3 and OpenAI Sora 2 Pro.
Deep ResearchAnysphere (Cursor) raised $2.3 billion Series D backed by Accel, Thrive, a16z, Nvidia, and Google. AI coding assistant reached $29.3B valuation.
Deep ResearchAnthropic releases Claude Opus 4.5, completing the Claude 4.5 family rollout with the most capable model in the series.
Web ResearchNVIDIA made first direct investment in Anthropic, committing up to $10 billion as part of strategic round that included $5 billion from Microsoft.
Web ResearchEuropean Commission issued Digital Omnibus proposals delaying certain AI Act obligations by up to 16 months to December 2027.
Deep ResearchGoogle launches Gemini 3 with focus on reasoning over text, code, images, and video. Gemini 3 Pro offers 1,048,576 token context window.
Web ResearchxAI releases Grok 4.1 with a massive 2-million-token context window, part of an unprecedented concentration of frontier model releases in late 2025.
Web ResearchAnthropic released Claude Opus 4.5 with 32k thinking tokens, achieving #1 on LMArena WebDev leaderboard and strong performance on SWE-bench Verified agentic coding benchmarks.
Web ResearchOpenAI released GPT-5.1, a smarter and more conversational ChatGPT version with improved natural language capabilities.
Deep ResearchGoogle secured DOJ clearance to acquire cloud security company Wiz for $32 billion, Google's largest acquisition ever, on track to close early 2026.
Deep ResearchAnthropic releases Claude 4.5 Sonnet, described as 'safety-first' with refined tone control and upgraded agentic/coding behavior.
Web ResearchGeneral Intuition scores a $134 million seed round to teach AI agents spatial reasoning for robotics and physical world interaction.
Web ResearchAnthropic launches Claude Code for web, an asynchronous coding agent enabling developers to delegate coding tasks to Claude.
Web ResearchMeta hired Andrew Tulloch from Thinking Machines Lab with compensation package rumored at $1.5 billion over six years, highlighting aggressive AI talent wars.
Deep ResearchOpenAI launches ChatGPT Atlas, a web browser built by a team including long-time Google Chrome engineers Ben Goodger and Darin Fisher.
Web ResearchGoogle released Veo 3.1 and Veo 3.1 Fast in paid preview, offering improvements in video quality and better outputs for image-to-video generation.
Deep ResearchNvidia became the first company to reach $5 trillion market cap, driven by unprecedented AI chip demand. Previously hit $4 trillion in July 2025.
Deep ResearchOpenAI and Sur Energy announced AI data center in Argentina's Patagonia, the first Stargate site in Latin America with up to $25 billion investment.
Deep ResearchIndia pilots AI chatbot-led e-commerce platform integrating ChatGPT, Gemini, and Claude, representing a major government-backed AI adoption initiative.
Web ResearchMira Murati's Thinking Machines Lab announced Tinker API for fine-tuning language models, the startup's first commercial product after $2B seed round at $12B valuation.
Deep ResearchOpenAI unveiled Sora 2 with iOS app, generating video with synchronized audio including dialogue, sound effects, and ambient soundscapes.
Deep ResearchAnthropic releases Claude 4.5 Haiku, a fast, cost-effective model optimized for high-volume, low-latency applications.
Web ResearchOpenAI, Oracle, and SoftBank announced five new U.S. AI data center sites under Stargate in Texas, New Mexico, and Ohio. Combined capacity nearly 7 gigawatts with $400B+ planned investment.
Deep ResearchModular announced $250 million funding led by US Innovative Technology Fund with GV, Greylock, and General Catalyst to advance its AI development platform.
Deep ResearchAI inference company Groq raised $750 million Series D-3 led by Disruptive, supporting its specialized chip technology for fast AI inference.
Deep ResearchAnthropic releases code execution sandbox feature, initially named 'Create and edit files with Claude', enabling safe code execution in isolated environments.
Web ResearchAlibaba released Qwen3-Next under Apache 2.0 license, including post-trained Instruct and Thinking models advancing open-source AI.
Deep ResearchAnthropic completed a $13 billion Series F round co-led by ICONIQ, Fidelity, and Lightspeed. Run-rate revenue reached over $5 billion by August 2025.
Deep ResearchxAI released Grok Code Fast 1, a specialized and cost-effective model for agentic coding that excels at automating software development workflows.
Deep ResearchGoogle launched Gemini 2.5 Flash Image (nicknamed 'Nano Banana'), representing advancement in image generation within the Gemini ecosystem.
Deep ResearchOpenAI launches a sub-$5 ChatGPT Go plan to attract more subscribers in India, expanding AI accessibility in emerging markets.
Web ResearchGoogle releases Gemini 2.5 Deep Think, the first publicly available multi-agent AI model that can spin up multiple reasoning agents in parallel to tackle different aspects of a problem.
Web ResearchxAI made unlimited access to Grok 4 available to free users for a limited time to compete with OpenAI's newly released GPT-5.
Deep ResearchOpenAI releases GPT-5, the much-anticipated successor setting the industry benchmark for complex reasoning, problem-solving, and code generation. Offers Auto, Fast, and Thinking modes.
Web ResearchAlibaba's Qwen team released Qwen-Image, followed by Qwen-Image-Edit on August 19th, expanding Qwen's multimodal capabilities beyond text.
Deep ResearchThe EU AI Act governance rules and obligations for General Purpose AI (GPAI) models became applicable, establishing strict safeguards and compliance requirements.
Deep ResearchxAI released Grok 4 with real-time reasoning and live data integration for conversational intelligence and contextual awareness.
Deep Research Q1-Q2Reka AI raised $110 million Series B led by Snowflake and Nvidia at $1 billion valuation for multimodal AI research.
Deep Research Q1-Q2Google launches AlphaEarth Foundations to help map our planet in unprecedented detail using AI-powered satellite imagery analysis.
Web ResearchMeta announces formation of an AI team to pursue superintelligence, reportedly offering nine-figure compensation packages to AI experts from competitors.
Web ResearchMira Murati's Thinking Machines Lab emerged from stealth with $2 billion at $12 billion valuation, led by a16z with Nvidia, AMD, Cisco.
Deep Research Q1-Q2OpenAI's $3B Windsurf acquisition collapsed due to Microsoft IP concerns. Google paid $2.4B for tech and CEO, Cognition acquired remaining assets.
Deep Research Q1-Q2OpenAI completed Jony Ive's io acquisition. io team merged with OpenAI, Ive and LoveFrom taking design responsibilities for AI hardware.
Deep Research Q1-Q2xAI launches Grok 4 with real-time reasoning and live data integration, targeting conversational intelligence and contextual awareness.
Web ResearchNVIDIA became first company to reach $4 trillion market cap, driven by overwhelming AI chip demand and Blackwell architecture dominance.
Deep Research Q1-Q2CoreWeave announced $9 billion acquisition of Core Scientific, expanding AI computing infrastructure in major pivot from crypto to AI workloads.
Deep Research Q1-Q2Zuckerberg announced Meta Superintelligence Labs with Alexandr Wang as Chief AI Officer, hiring 11 engineers from Anthropic, DeepMind, OpenAI with $100M+ packages.
Deep Research Q1-Q2DeepMind released AlphaGenome to interpret genomic 'dark matter', handling 1 million base-pair sequences and predicting gene expression for drug discovery.
Deep Research Q1-Q2Google announced general availability for Gemini 2.5 Pro and Flash, making advanced thinking models widely accessible through AI Studio and Vertex AI.
Deep Research Q1-Q2Helsing, a European developer of AI autonomous defence technology, recruits Saab for a $691 million funding round.
Web ResearchMistral AI released first reasoning models: Magistral Small (open-source) and Medium with chain-of-thought for analytical problem-solving.
Deep Research Q1-Q2Survey found 61% of American adults used AI tools in past six months, representing 1.8 billion users globally with 500-600 million daily.
Deep Research Q1-Q2Cursor, founded by recent MIT graduates, raises $900 million at a $10 billion valuation. The company went from zero to $100 million in recurring revenue in less than two years.
Web ResearchAnthropic released Claude Opus 4 and Sonnet 4 setting new standards for coding. Opus 4 leads SWE-bench (72.5%) and classified as Level 3 high-risk model.
Deep Research Q1-Q2OpenAI announced acquisition of Jony Ive's io hardware startup for $6.5 billion all-stock, bringing 55 hardware engineers for AI consumer devices.
Deep Research Q1-Q2Google launches Jules, an asynchronous coding agent, along with Stitch for UI design and Google Beam, an AI-first 3D video communication platform.
Web ResearchGoogle unveiled Veo 3 (video with audio and dialogue), Imagen 4 (2K images), and Flow (AI filmmaking) at I/O 2025.
Deep Research Q1-Q2Google releases Veo 3 video generation, Imagen 4 image generation, and Flow, new generative media models and tools for creative applications.
Web ResearchAI infrastructure company TensorWave raised $100 million Series A co-led by Magnetar Capital and AMD Ventures.
Deep Research Q1-Q2Anthropic launches the Claude 4 family including Claude Opus 4 and Claude Sonnet 4, featuring hybrid architecture with instant responses and extended thinking capabilities for chain-of-thought reasoning.
Web ResearchMistral AI released Mistral Medium 3, enterprise-grade model balancing performance with cost efficiency.
Deep Research Q1-Q2OpenAI announced plans to acquire AI coding IDE Windsurf for $3 billion, its largest acquisition. Later fell apart due to Microsoft IP concerns.
Deep Research Q1-Q2At Cloud Next 25, Google unveiled Agent2Agent (A2A) open protocol for AI agent collaboration across frameworks, plus Ironwood TPU.
Deep Research Q1-Q2OpenAI released o3 scoring 87.7% on GPQA Diamond and 71.7% on SWE-bench, plus o4-mini optimized for fast, cost-efficient reasoning.
Deep Research Q1-Q2OpenAI releases GPT-4.1 featuring a significantly larger context window, improving long-form content handling and complex task completion.
Web ResearchMeta releases Llama 4 with native multimodal capabilities. The launch was controversial due to discrepancies between the model tested on LMArena and the released version.
Web ResearchMeta launched Llama 4 with Scout (17B, 10M context) and Maverick (400B, multimodal) using mixture-of-experts, beating GPT-4o and Gemini 2.0 Flash.
Deep Research Q1-Q2GitHub rolled out Copilot Agent Mode with MCP support to all VS Code users, enabling AI to iterate on code and fix mistakes in real-time.
Deep Research Q1-Q2OpenAI announced $40 billion investment led by SoftBank, the largest venture investment ever, potentially valuing OpenAI at $830 billion.
Deep Research Q1-Q2Google launched Gemini 2.5 Pro Experimental with built-in thinking and 1M token context, debuting #1 on LMArena with 18.8% on Humanity's Last Exam.
Deep Research Q1-Q2OpenAI launched native image generation within GPT-4o enabling precise text rendering, photorealistic outputs, and iterative refinement through conversation.
Deep Research Q1-Q2Google announced agreement to acquire cloud security startup Wiz for $32 billion all-cash, Alphabet's largest acquisition ever.
Deep Research Q1-Q2At GTC 2025, NVIDIA revealed Blackwell Ultra chips with 288GB HBM3e shipping H2 2025, and previewed Vera Rubin GPU for 2026.
Deep Research Q1-Q2Mistral AI released Mistral Small 3.1, a more efficient version continuing their strategy of cost-effective alternatives to larger models.
Deep Research Q1-Q2AI evaluation organization METR releases research showing that every seven months, the length of software engineering tasks that leading AI models could complete was doubling.
Web ResearchGoogle announces Gemini 2.5, described as its 'most intelligent model yet' with enhanced reasoning and coding skills, along with native image generation capabilities.
Web ResearchGoogle unveiled Gemma 3 open-source multimodal model in 1B, 4B, 12B, 27B sizes with 128K context, 35+ languages, outperforming larger models.
Deep Research Q1-Q2ServiceNow announced largest acquisition ever, buying AI assistant company Moveworks for $2.85 billion in cash and stock.
Deep Research Q1-Q2Anthropic raised $3.5 billion Series E led by Lightspeed Venture Partners at $61.5 billion valuation for next-gen AI development and international growth.
Deep Research Q1-Q2HP announced acquisition of key AI capabilities from Humane, including Cosmos operating platform and 300+ patents related to AI wearables and assistants.
Deep Research Q1-Q2OpenAI adopted Anthropic's Model Context Protocol (MCP) across Agents SDK, Responses API, and ChatGPT desktop. Sam Altman: 'People love MCP and we are excited to add support.'
Deep Research Q1-Q2OpenAI released GPT-4.5 (codenamed Orion), its largest model with estimated 4-5 trillion parameters, featuring improved pattern recognition and reduced hallucinations.
Deep Research Q1-Q2IBM completed $6.4 billion acquisition of HashiCorp, integrating Terraform and Vault into IBM's hybrid cloud and automation portfolio.
Deep Research Q1-Q2Microsoft launched Phi-4-mini (3.8B) with 128K context and Phi-4-multimodal (5.6B) processing speech, vision, and text, achieving top positions on speech benchmarks.
Deep Research Q1-Q2Anthropic launched Claude 3.7 Sonnet, the first hybrid reasoning model with extended thinking mode, along with Claude Code agentic command-line coding tool achieving state-of-the-art on SWE-bench.
Deep Research Q1-Q2Startup Inception Labs releases Mercury, an innovative text diffusion model aimed at coding tasks, representing a new approach to code generation.
Web ResearchxAI releases Grok 3, trained on the massive Colossus supercomputer with 200,000+ NVIDIA H100 GPUs. The model emphasizes truth-seeking AI with powerful reasoning capabilities.
Web ResearchAI hardware startup EnCharge AI closed $100 million Series B led by Tiger Global for specialized AI hardware for efficient model inference.
Deep Research Q1-Q2Anthropic releases Claude 3.7 Sonnet along with Claude Code, a revolutionary coding assistant that would become one of the most impactful AI tools of 2025.
Web ResearchMistral AI released Le Chat on iOS and Android along with a Pro subscription at $14.99/month, expanding consumer-facing AI assistant availability.
Deep Research Q1-Q2The EU AI Act's prohibitions on 'unacceptable risk' AI practices became legally binding across all 27 EU member states, banning social scoring, emotion recognition in workplaces, and untargeted facial recognition.
Deep Research Q1-Q2OpenAI releases o3-mini reasoning model to all users including free tier, matching o1 performance with significantly improved coding abilities.
AI TimelinesOpenAI released o3-mini with three effort levels (low, medium, high) for cost-efficient reasoning using chain-of-thought techniques for math, coding, and science.
Deep Research Q1-Q2DeepSeek R1 surpasses ChatGPT as the most downloaded freeware app on iOS App Store in the US, causing 18% drop in Nvidia stock.
AI TimelinesOpenAI released Operator, an AI agent capable of autonomously performing web-based tasks like booking travel, shopping, and filling forms, powered by the Computer-Using Agent (CUA) model.
Deep Research Q1-Q2OpenAI releases Operator, an experimental AI agent capable of browsing websites and performing actions autonomously.
AI TimelinesPresident Trump announced the Stargate project with OpenAI, SoftBank, and Oracle, committing up to $500 billion for AI infrastructure development in the US, aiming to build 20 data centers and create 100,000 jobs.
Deep Research Q1-Q2DeepSeek releases R1 reasoning model with 671B parameters, open-sourced under MIT license, claiming only $6M training cost and matching o1 performance.
AI TimelinesGoogle releases Gemini Flash Thinking 0121, an enhanced reasoning model securing top spot in Arena Chatbots rankings.
AI TimelinesDeepSeek releases Janus Pro 7B, a fully open-sourced multimodal model supporting text and image generation.
AI TimelinesAlibaba releases Qwen2.5-Max LLM surpassing DeepSeek-V3, GPT-4o, and Claude 3.5, plus Qwen2.5-1M supporting 1M token context.
AI TimelinesNvidia completed its acquisition of Israeli AI infrastructure startup Run:ai for approximately $700 million. As part of the deal, Run:ai's software will be open sourced.
TechCrunchxAI raised $6 billion in a Series C round, bringing total 2024 fundraising to $12 billion. Investors included Andreessen Horowitz, Blackrock, Fidelity, and Nvidia, valuing the company at $45 billion.
TechCrunchAI search engine Perplexity closed a $500 million funding round led by Institutional Venture Partners, valuing the startup at $9 billion. The company raised multiple rounds in 2024, growing from a $520M valuation in January.
TechCrunchAI coding assistant Cursor raised another $100 million at a $2.6 billion valuation led by Thrive Capital, just four months after its previous funding round.
TechCrunchAI search engine Perplexity acquired Seattle-based startup Carbon, which specializes in connecting AI systems to external data sources using retrieval augmented generation (RAG).
TechCrunchDatabricks raised $10 billion at a $62 billion valuation, making it the largest AI funding round of 2024, surpassing even OpenAI's $6.6B round. The round was backed by Thrive Capital, Andreessen Horowitz, DST Global, and others.
TechCrunchLiquid AI secured substantial funding to continue developing efficient, general-purpose liquid-style foundation models.
TLDR AIThe AI researcher launched World Labs focusing on spatial intelligence through 3D scene creation for robotics, VR, and AR applications.
TLDR AIMicrosoft released a synthetically trained model excelling on challenging benchmarks, confirming mid-training effectiveness for context extension.
TLDR AIMeta's token-free training method uses learned patches instead of tokens, demonstrating superior scaling on smaller models.
TLDR AIGoogle's Willow quantum chip significantly reduces qubit error rates, advancing quantum computing capabilities.
TLDR AIMicrosoft released software converting Office files (docx, xlsx, ppt) to markdown for use as LLM context.
TLDR AIGoogle releases Veo 2 beta video generation model capable of producing 4K videos up to two minutes, outperforming Sora in human evaluations.
AI TimelinesDeepSeek releases V3, an open-source model with 671B parameters surpassing several closed-source benchmarks.
AI TimelinesAmazon introduces Nova model series for text, image, and video processing capabilities.
AI TimelinesMeta releases Llama 3.3 70B offering performance comparable to Llama 3.1 405B.
AI TimelinesMicrosoft open-sources Phi-4, a 14B parameter small language model demonstrating strong performance for its size.
AI TimelinesxAI releases Aurora, a new image generation model integrated into their systems.
AI TimelinesAlibaba releases QVQ-72B-Preview, a thinking model capable of image analysis with state-of-the-art performance.
AI TimelinesGoogle releases PaliGemma 2, a multimodal open-source model integrated with existing Gemma models.
AI TimelinesMeta releases Apollo video generation model available in three different sizes.
AI TimelinesAnthropic released Claude 3.5 Haiku, which matches or exceeds the performance of the previous flagship Claude 3 Opus on specific benchmarks, particularly for coding and data extraction tasks.
TechCrunchCharacter AI unveiled new safety features for teenage users including content filters and usage notifications following multiple lawsuits accusing the company of harm to minors.
TechCrunchGoogle launches Gemini 2.0 Flash in beta with live speech and video capabilities, outperforming Gemini 1.5 Pro.
AI TimelinesAudio AI startup founded by Alexis Conneau (who developed ChatGPT's Advanced Voice Mode) secured funding to focus on solving the Turing test for audio AI systems.
TLDR AIVideo generation model Sora now available to Pro subscribers, marking expanded access to OpenAI's generative video technology.
TLDR AIDeepMind's GenCast weather forecasting system outperforms traditional ECMWF methods.
TLDR AIOpenAI releases Sora video generation model to ChatGPT Pro and Plus users in the US and Canada.
AI TimelinesOpenAI previews o3 reasoning model achieving 87.5% on ARC-AGI benchmark and 25.2% on Frontier Math Benchmark.
AI TimelinesMeta released Llama 3.3 70B, claiming it delivers the performance of the 405B model while being more cost-efficient to run. By year-end, Meta AI had nearly 600 million monthly active users.
TechCrunchMeta is collaborating with nuclear energy developers to add 1-4 gigawatts of nuclear capacity in the U.S., aimed at powering their AI innovation efforts sustainably.
TLDR AIOpenAI began hosting 12 consecutive live streams to unveil new product features and model capabilities, coinciding with ChatGPT's second-year anniversary.
TLDR AIGoogle DeepMind's large-scale latent diffusion model generates interactive worlds with character control inputs while maintaining consistency and visual quality.
TLDR AIOpenAI releases the full o1 reasoning model with image analysis, available to Plus and Pro subscribers.
AI TimelinesOpenAI launches ChatGPT Pro subscription at $200/month with unlimited access to o1 Pro mode.
AI TimelinesAnthropic raised an additional $4 billion from Amazon, bringing Amazon's total investment to $8 billion. The deal made AWS Anthropic's primary cloud partner for AI model training.
TechCrunchDeepSeek unveiled its r1 'Deep Think' model, replicating capabilities of OpenAI's o1. The company plans full open-source release of this reinforcement-learning-trained system.
TLDR AIResearch demonstrated that ChatGPT-4 exceeded physician performance in diagnosing medical conditions, though doctors often disregarded AI insights in favor of personal judgment.
TLDR AIGoogle DeepMind developed a model identifying and correcting quantum computation errors, advancing quantum error correction capabilities.
TLDR AIThe creator of Keras and ARC eval departed Google to pursue new projects while continuing to support the Jax and Keras communities.
TLDR AIOpenAI unveiled infrastructure proposals including AI economic zones and Navy partnerships on nuclear power, aiming to boost innovation and maintain competitive advantage over China.
TLDR AIBuilt on Qwen 2.5 72B, Athene-V2 models are optimized for agentic and chat workflows, outperforming GPT-4o on multiple critical benchmarks.
TLDR AIAnthropic releases Claude 3.5 Haiku alongside visual PDF analysis feature.
AI TimelinesAnthropic announces Model Context Protocol (MCP), a universal interface standard for connecting AI models with external tools and data sources.
AI TimelinesAlibaba releases QwQ 32B Preview reasoning model that competes with and sometimes surpasses OpenAI's o1-preview.
AI TimelinesDeepSeek releases R1-Lite-Preview reasoning model with AIME and MATH performance matching o1-preview levels.
AI TimelinesAlibaba releases Qwen2.5 Coder 32B, an open-source coding model with capabilities comparable to leading proprietary models.
AI TimelinesMistral releases Pixtral Large, a multimodal language model excelling in image recognition.
AI TimelinesGoogle released a dedicated Gemini app for iOS globally, following the February launch on Android. The app brought Google's AI assistant to iPhone users worldwide.
TechCrunchMeta announced partnerships with Accenture, AWS, Palantir, Lockheed Martin, and others to bring Llama models to US government agencies and defense contractors.
TechCrunchAnthropic launched Claude desktop apps for Mac and Windows in public beta, along with dictation support allowing users to record and upload voice messages up to 10 minutes.
TechCrunchGitHub announced that Copilot Chat would allow developers to switch between multiple LLMs including Anthropic's Claude 3.5 Sonnet, Google's Gemini 1.5 Pro, and OpenAI's GPT-4o.
TechCrunchAnthropic enhanced Claude 3.5 Sonnet with the ability to interact with computers, trained to use computers and click on things using screen recordings.
TLDR AIStability AI unveiled an improved model series with enhanced performance and faster processing, offering built-in Diffusers integration.
TLDR AINews Corp sued AI search engine Perplexity for 'content kleptocracy,' alleging unauthorized scraping of news articles. The New York Times also sent a cease and desist letter to Perplexity.
TechCrunchAdobe expanded its Firefly generative AI family to include video capabilities, announced during Adobe MAX conference.
TLDR AIOpenAI appointed Oliver Jay to lead international expansion, opened a new APAC office in Singapore, and is developing local language datasets.
TLDR AIAnthropic releases upgraded Claude 3.5 Sonnet with significant improvements and introduces Computer Use beta for AI-controlled desktop interaction.
AI TimelinesMeta announces Movie Gen, an AI model capable of generating videos, images, and audio from text input.
AI TimelinesOpenAI officially launches SearchGPT web search capability, allowing users to perform web searches directly in ChatGPT.
AI TimelinesStability AI releases Stable Diffusion 3.5 in three sizes as open source.
AI TimelinesAdobe announces Firefly Video, their AI video creation model.
AI TimelinesMistral releases Ministral open-source models in 3B and 8B parameter sizes.
AI TimelinesBlack Forest Labs releases Flux 1.1 Pro image creation model with advanced capabilities.
AI TimelinesOpenAI closed the largest venture round of all time at $6.6 billion, led by Thrive Capital with participation from Microsoft, Nvidia, SoftBank, and others. The company also secured a $4 billion credit line.
TechCrunchPoolside, an AI-powered software development platform, closed a $500 million Series B round led by Bain Capital Ventures with participation from Nvidia, valuing the company at $3 billion.
TechCrunchAt Meta Connect 2024, Meta announced Llama 3.2 with 11B and 90B multimodal models that can interpret charts, caption images, and pinpoint objects in pictures.
TechCrunchMeta releases Llama 3.2 in 1B, 3B, 11B, and 90B sizes, featuring image recognition capabilities for the first time.
AI TimelinesRunway became the first AI startup to publicly partner with a major Hollywood studio, training a custom video model on Lionsgate's catalog including 'John Wick' and 'Twilight' franchises.
TechCrunchOpenAI plans restructuring next year, claiming it has outgrown its current structure and aims to simplify operations while becoming more appealing to investors.
TLDR AISalesforce launched autonomous AI agents capable of independent action within defined operational boundaries.
TLDR AIOpenAI launches Advanced Voice Mode for GPT-4o, enabling natural real-time voice conversations.
AI TimelinesAlibaba releases Qwen 2.5 model series from 0.5B to 72B parameters with capabilities comparable to much larger models.
AI TimelinesMistral releases Pixtral 12B, their first multimodal model capable of processing images and text.
AI TimelinesOpenAI released o1, a reasoning-focused model trained to think before it answers, demonstrating superhuman performance in certain domains. Usage initially constrained to ~30 queries per user weekly.
TLDR AIGoogle deployed Gemini Live to free Android users with conversational AI features allowing users to interrupt responses and access text transcripts.
TLDR AIMistral launched Pixtral 12B, a 12-billion-parameter multimodal model capable of processing both images and text, available under Apache 2.0 licensing.
TLDR AIOpenAI is in talks to raise $6.5 billion at a $150 billion valuation.
TLDR AIOpenAI releases o1-preview and o1-mini reasoning models that 'think' before responding, excelling at complex reasoning, coding, and mathematics.
AI TimelinesSafe Superintelligence (SSI), founded by former OpenAI chief scientist Ilya Sutskever, raised over $1 billion from investors including a16z, Sequoia, and DST Global at a $5 billion valuation.
TechCrunchAnthropic launched Claude Enterprise with a 500,000 token context window (200,000 lines of code), GitHub integration, and enhanced security features to compete with ChatGPT Enterprise.
TechCrunchGenerative AI coding startup Magic raised $320 million in Series C funding from Eric Schmidt, CapitalG, Atlassian, and Sequoia. The company develops AI tools with ultra-long context windows for software development.
TechCrunchxAI unveiled Grok 2, a frontier-class model with capabilities in reasoning, code, and mathematics. The company partnered with Black Forest Labs to integrate FLUX into X's platform.
TLDR AIAnthropic launched prompt caching for Claude models, enabling developers to cache frequently used context, reducing costs and latency.
TLDR AINvidia released Llama 3.1 MiniTron 4B, a compressed version achieving 16% better MMLU scores than training from scratch using knowledge distillation and pruning while requiring 40x fewer tokens.
TLDR AIMIT launched a comprehensive database documenting over 700 AI risks with classification systems for organizational assessment.
TLDR AIBlack Forest Labs releases Flux, an image creation model with open weights outperforming similar closed-source models.
AI TimelinesxAI releases Grok 2 and Grok 2 Mini with performance on par with leading state-of-the-art models.
AI TimelinesMicrosoft releases Phi 3.5 in three versions, small language models with impressive performance relative to size.
AI TimelinesGroq raised $640 million led by Blackrock to develop chips that run AI models faster than conventional processors. The round valued the company at $2.8 billion and brought total funding over $1 billion.
TechCrunchCharacter.AI co-founder and CEO Noam Shazeer returned to Google along with co-founder Daniel De Freitas. Google signed a non-exclusive licensing agreement with Character.AI in a deal valued at $2.7 billion.
TechCrunchThe European Union's risk-based AI regulation became law, establishing global standards for AI governance. The law bans certain AI applications and sets compliance deadlines from 2025 to 2027.
TechCrunchMeta released its first open frontier AI model with comprehensive training details, featuring 128k context length, chat variants, and performance parity with leading proprietary systems.
TLDR AIThe legal technology AI startup achieved unicorn status, securing funding from Google Ventures to expand adoption among major law firms.
TLDR AIMark Zuckerberg articulated Meta's commitment to open source as fundamental to its AI strategy and outlined the company's planned support mechanisms.
TLDR AICanadian AI startup Cohere achieved a $5.5B valuation through a $500M Series D funding round, supported by PSP Investments, Cisco, Fujitsu, AMD Ventures, and EDC.
TLDR AIMeta releases Llama 3.1 in 8B, 70B, and 405B parameter sizes, with the 405B model matching closed-source capabilities.
AI TimelinesEnterprise AI startup Cohere raised $500 million from Cisco, AMD, and Fujitsu, valuing the Toronto-based company at $5.5 billion. The company focuses on secure enterprise AI solutions.
TechCrunchOpenAI releases GPT-4o Mini, offering high capabilities at significantly lower cost than GPT-4o.
AI TimelinesMenlo Ventures and Anthropic launched the Anthology Fund, a $100 million initiative to invest in pre-seed and Series A AI startups building on Anthropic's Claude technology.
TechCrunchOpenAI is developing advanced reasoning capabilities under the secret project 'Strawberry,' aiming to enable AI models to autonomously research the internet and plan for complex tasks.
TLDR AIOpenAI blocked users in China from accessing its tools and services, closing a workaround that allowed developers to use VPNs.
TLDR AIOpenAI's estimated revenue reaches $3.4 billion, predominantly from ChatGPT services.
TLDR AIGoogle DeepMind's AlphaProof and AlphaGeometry 2 win silver medals at the International Mathematical Olympiad.
AI TimelinesMistral AI releases Mistral Large 2, a 123B parameter model with capabilities close to closed SOTA models.
AI TimelinesOpenAI announces SearchGPT, an integrated web search capability for ChatGPT.
AI TimelinesAmazon hired the co-founders and key team members from Adept, an AI agent startup that had raised over $415 million at a $1 billion valuation. Amazon also licensed Adept's technology.
TechCrunchAnthropic released Claude 3.5 Sonnet, which passes Opus in performance at 1/5 the cost. Described as the best vision model available at the time.
TLDR AIApple contributed 20 Core Machine Learning models to Hugging Face, enhancing image classification and depth segmentation capabilities.
TLDR AIFactory secured $15M in Series A funding led by Sequoia Capital. Their AI-powered software development tools achieved state-of-the-art results on the SWE-bench coding benchmark.
TLDR AIAnthropic releases Claude 3.5 Sonnet, outperforming Claude 3 Opus at twice the speed with new Artifacts feature for real-time code previews.
AI TimelinesOpenAI co-founder Ilya Sutskever announced Safe Superintelligence (SSI), focused solely on building safe superintelligent AI. The startup has offices in Palo Alto and Tel Aviv.
TechCrunchNVIDIA's Nemotron-4 340B is a family of open models that developers can use to generate synthetic data for training LLMs. The reward model matches GPT-4 performance.
TLDR AIApple won't pay OpenAI directly for ChatGPT integration; the company believes distributing OpenAI's technology across its devices provides equivalent value.
TLDR AIAmazon commits up to $230 million to support startups building generative AI-powered applications.
TLDR AIDeepSeek releases open-source DeepSeekCoder V2 with coding capabilities comparable to GPT-4 and Claude 3 Opus.
AI TimelinesRunway releases Gen3 Alpha, their latest AI video generation model.
AI TimelinesGoogle announces Gemma 2 open-source language models in 9B and 27B parameter sizes.
AI TimelinesStability AI releases Stable Diffusion 3 Medium, a 2B parameter image creation model.
AI TimelinesFrench AI startup Mistral AI raised 600 million euros ($640M) in Series B funding led by General Catalyst, valuing the company at $6 billion. Investors included Lightspeed, a16z, Nvidia, and IBM.
TechCrunchApple unveils Apple Intelligence at WWDC, integrating AI across iOS, macOS with on-device and cloud processing.
AI TimelinesApple announced Apple Intelligence, its ecosystem-wide push into generative AI featuring a revamped Siri, ChatGPT integration, Genmoji, Image Playground, and new writing tools across iOS 18, iPadOS 18, and macOS Sequoia.
TechCrunchData-labeling startup Scale AI raised $1 billion in a Series F round from Accel, Amazon, and Meta, doubling its valuation to $13.8 billion. The company serves major clients including Microsoft, OpenAI, and the US Department of Defense.
TechCrunchGoogle unveiled Gemini Flash, Veo video generation, Imagen 3, and Project Astra featuring 2M token context length, dramatically cheaper models, and improved multimodality.
TLDR AIAnthropic rolled out Claude across Europe with multilingual support via website, iOS app, and team business plans while beginning new fundraising efforts.
TLDR AIxAI and Oracle are in discussions for a potential $10 billion cloud server rental agreement.
TLDR AIGoogle releases Gemini 1.5 Flash, a smaller and faster model with 2 million token context window.
AI TimelinesMicrosoft announces Copilot+ PCs and releases Phi-3 Small, Phi-3 Medium, and Phi-3 Vision models as open source.
AI TimelinesMistral AI releases Codestral, a 22B parameter model designed specifically for code generation.
AI TimelinesGoogle unveiled Gemini 1.5 Pro, Project Astra for AI agents, Gemini Live for voice chats, Trillium TPU chips, and Ask Photos feature powered by Gemini at its annual developer conference.
TechCrunchGPT-4o integrates vision and voice to become a more ubiquitous presence in daily life, marking OpenAI's multimodal advances.
TLDR AIMeta releases Llama 3 open-source models in 8B and 70B parameter sizes, outperforming Claude 3 Sonnet and Gemini Pro 1.5 on most benchmarks.
AI TimelinesxAI announced its latest flagship model features vision capabilities that match or exceed state-of-the-art competitors.
TLDR AIGoogle released Cloud TPU v5p offering nearly triple the training speed for LLMs compared to its predecessor, and introduced the Google Axion CPU based on Arm architecture.
TLDR AIMicrosoft releases Phi-3-mini, a 3.8B parameter open-source model capable of running on mobile devices with GPT-3.5-like capabilities.
AI TimelinesxAI releases Grok-1.5V with high-level image recognition capabilities, excelling at identifying and analyzing images.
AI TimelinesApple releases OpenELM, a series of small language models as full open source in four sizes (270M to 3B parameters).
AI TimelinesAdobe announces Firefly 3, their latest image creation model with improved capabilities.
AI TimelinesMistral AI launches Mixtral 8x22B, a sparse MoE model with 141B total parameters and 39B active parameters.
AI TimelinesSam Altman and Jony Ive are collaborating to create an AI-powered personal device that won't resemble a smartphone, seeking up to $1 billion in funding.
TLDR AIPrime Minister Justin Trudeau announced a $2.4 billion investment package including funding for computing infrastructure, AI startups, a Canadian AI Safety Institute, and enforcement of the AI and Data Act.
TLDR AILightweight LLM training implementation in C/CUDA, eliminating heavy PyTorch dependencies.
TLDR AIAI21 Labs unveils Jamba, a hybrid Mamba-style model with 256K context window and efficient GPU deployment.
AI TimelinesStability AI founder Emad Mostaque stepped down as CEO and board member, citing desire to pursue decentralized AI. The company appointed its COO and CTO as interim co-CEOs amid financial troubles.
TechCrunchMicrosoft hired Inflection AI's co-founders Mustafa Suleyman and Karen Simonyan to run a new consumer AI division. The deal was valued at approximately $650 million in licensing fees and payments.
TechCrunchThe European Parliament approved a comprehensive AI regulatory framework establishing global standards, risk categories, and implementation beginning in 2025.
TLDR AIFigure implemented OpenAI's technology, combining visual and language models with the robot's neural networks for dexterous task execution and dialogue.
TLDR AISuno releases v3 of their AI music creation model to the general public.
AI TimelinesApple acquired Canada-based DarwinAI, which specializes in vision-based technology for manufacturing efficiency. The startup had raised over $15 million from investors including Honeywell Ventures.
TechCrunchThe European Parliament gave final approval to the EU AI Act following affirmative committee votes, making it the world's first comprehensive AI regulation framework.
TechCrunchInflection AI releases Inflection-2.5, achieving 94% of GPT-4 performance using only 40% of training compute.
AI TimelinesAnthropic released three new Claude 3 models, with the strongest (Opus) matching GPT-4's benchmark performance. The release includes multimodal capabilities and substantially improved coding ability.
TLDR AIMusk claims GPT-4 constitutes AGI and is no longer subject to the Microsoft agreement, demanding greater transparency and openness from the company.
TLDR AIFigure AI secured $675 million in funding and partnered with OpenAI to enhance humanoid robot capabilities through improved semantic understanding.
TLDR AIAnthropic releases Claude 3 family (Haiku, Sonnet, Opus) with Claude 3 Opus outperforming GPT-4 on most benchmarks.
AI TimelinesFigure AI raised $675 million in Series B funding from Microsoft, OpenAI Startup Fund, Nvidia, Amazon, and Jeff Bezos to develop humanoid robots. The round valued the company at $2.6 billion.
TechCrunchGoogle released weights for Gemma 2B and 7B parameter models via HuggingFace, substantially exceeding Llama 2 performance across multiple benchmarks.
TLDR AIReddit entered a licensing agreement worth approximately $60 million annually, granting a major AI company access to user-generated content for model training.
TLDR AIOpenAI unveiled Sora, a new video generation model with stunning quality that is available for red teamers to test.
TLDR AILambda Labs secured $320M in Series C funding to expand its GPU cloud infrastructure for AI applications.
TLDR AIMeta released V-JEPA, a video-based self-supervised model developed by Chief AI Scientist Yann LeCun, aimed at improving data efficiency in AI systems.
TLDR AIOpenAI founding member and former Tesla Autopilot leader announced his departure to focus on personal projects.
TLDR AINVIDIA launched Chat with RTX, enabling local analysis and summarization of personal data without cloud processing, currently limited to Windows with specific GPUs.
TLDR AIGoogle developed a new text-to-video model accepting images and styles as input, using a novel 'space-time UNet' for simultaneous diffusion.
TLDR AIOpenAI announces Sora, a groundbreaking text-to-video model capable of generating realistic videos up to one minute long.
AI TimelinesGoogle announces Gemini 1.5 Pro with up to 1 million token context window, using Mixture-of-Experts architecture.
AI TimelinesStability AI announces Stable Diffusion 3 with gradual rollout to waitlist users.
AI TimelinesAbacus AI developed Smaug-72B, derived from Alibaba's Qwen-72B, now topping Hugging Face's NLP leaderboard, outperforming GPT-3.5 and Mistral Medium with an average score of over 80 across major evaluations.
TLDR AIThe European Union's AI Act, focused on risk-based regulation, passed a crucial vote by Member State representatives confirming the final draft text.
TLDR AIAI hardware device achieving $10M in preorders through collaborative design between Rabbit and Teenage Engineering.
TLDR AIGoogle's Bard AI chatbot officially rebrands to Gemini, reflecting the underlying model name.
AI TimelinesOpenAI releases text-embedding-3-small and text-embedding-3-large models with improved performance.
AI TimelinesStability AI releases Stable LM 2 language models with up to 12 billion parameters.
AI TimelinesOpenAI revised its usage policy to potentially allow military applications of its technologies, marking a departure from previous restrictions.
TLDR AIMIT researchers employed deep learning to discover compounds that can kill drug-resistant MRSA bacteria, responsible for over 10,000 annual U.S. deaths.
TLDR AIAnthropic's research revealed that 'poisoned' language models cannot be realigned with current safety tools, even after specific training interventions.
TLDR AIOpenAI launches GPT Store for discovering and sharing custom GPTs, alongside ChatGPT Team subscription for businesses.
AI TimelinesApple plans to debut AI-powered features at its June developer conference, including an enhanced Siri based on the Ajax language model with capabilities for auto-completion, summarization, and Xcode coding assistance.
TLDR AIOpenAI claimed the New York Times' copyright lawsuit is 'without merit,' arguing that training AI on publicly available data is fair use. The Times sued in December 2023 after licensing negotiations failed.
TechCrunchTLDR AI, AI Timelines, TechCrunch
Coverage: 2024-01-08 ~ 2026-02-21