Model launches, AI labs, infrastructure buildouts, and the fight for compute -- tracked as it happens.
Chinese collaborative-robot maker Faolf Robotics closed back-to-back Series C+ and Series D rounds totaling nearly $140 million, led again by delivery giant Meituan, as it pushes into humanoid force-control arms.
CoreWeave's stock and bonds sold off after investors focused on the roughly $52 billion in debt and leases behind its AI data centers, reviving doubts about who actually pays if AI demand cools.
Cal AI, founded by 19-year-old Zach Yadegari in 2024, grew past $30M in revenue through paid influencer marketing before MyFitnessPal acquired it in March; Yadegari then raised $10M for new startup Persona.
Meta released Muse Spark Open, an 816-billion-parameter open-weight model with a 1-million-token context window built to run locally on Apple's new Mac Studio with the M5 Ultra chip.
Anthropic is giving six researchers from AI-safety nonprofit METR company laptops, office access and internal tools on par with its own risk teams, and has given up the right to block their findings from publication.
Bluesky expanded its AI-powered Attie tool so anyone can build websites, quizzes, games and polls on the AT Protocol, not just custom social feeds, as the network looks to developers to counter a recent engagement slide.
Accenture's global AI CTO Marc Warner says almost no large enterprises have been truly reinvented by AI, even as AXA narrows about 400 pilots into a push for €500-700M in annual AI value by 2029.
Volvo is expanding its decade-long Nvidia partnership with three new components -- Hyperion, Alpamayo and Halos -- aiming for an automated driving system that works on both city streets and highways, TechCrunch reported.
Anthropic reportedly posted $16.8 billion in third-quarter revenue, but the pace of growth is decelerating sharply from the prior quarter -- a trend that could temper IPO valuation expectations near $2 trillion.
Google launched Universal Cart in the US on October 15, letting shoppers save and track products across Search and Gemini with price-drop alerts, directly challenging agentic-commerce pushes from Amazon and OpenAI.
Rivian CEO RJ Scaringe says future vehicles will run AI agents from outside partners, citing Anthropic as an example, with more details coming at a December event.
Cohere signed a letter of intent to buy AISLE, a San Francisco AI startup that hunts software vulnerabilities, adding security to the AI lab's pending $20 billion merger with Aleph Alpha.
Opsin raised $21 million led by SYN Ventures and Race Capital to give enterprises visibility into employee-built AI agents, after finding 93% of misaligned agents had unrestricted system access.
CedarDB launched CedarDB Cloud in limited availability at TechCrunch Disrupt, betting that AI agents' mixed, unpredictable query patterns will force companies off separate transactional and analytical databases and onto one PostgreSQL-compatible engine.
Colossal Biosciences unveiled GenoMorph, an AI platform combining genomics and developmental biology that has already flagged about 2,300 candidate gene edits for reviving the dodo's skull shape.
Sydney-based Apate deploys AI agents that pose as victims to draw out scammers, then sells the resulting fraud intelligence -- bank accounts, phone numbers, phishing sites -- to banks and telecoms.
Nvidia and ChipAgents are launching a joint evaluation of AI-powered chip design tools, extending a partnership that began with a $60 million Series A ChipAgents raised in July.
Google is giving 200 European researchers selected under the EU's RAISE program free access to frontier AI tools including AlphaFold 3 and WeatherNext, announced at a Dublin AI summit.
Nvidia will donate compute valued at $250 million over five years to fund agentic AI safety and security research, with the majority going to the wider UK research community and portions to Oxford's BOLD lab, UCL's SOFAIR lab, and Imperial College London.
Industrial Next unveiled AI-driven robotic stations that retool factory lines in minutes instead of weeks at TechCrunch Disrupt 2026, alongside a new partnership with BMW Group Technology Office.
Cerebras CEO Andrew Feldman told TechCrunch Disrupt that AI leaders' public doomsday warnings are overshadowing the technology's potential benefits, arguing some executives 'shouldn't be allowed to talk in public.'
Google released Gemini 4 Argon globally on Oct. 13, pricing API access at $2/$10 per million tokens through Nov. 30 before the rate doubles to $4/$20.
A CEO affidavit filed in Character.AI's Chapter 11 case shows the company burning $8 million to $9 million a month as of August, and describes Disney's much-reported $20 million as a licensing offer, not a confirmed asset sale.
Character.AI filed for Chapter 11 bankruptcy five years after launch, with Disney agreeing to buy some of its assets for $20 million as the AI chatbot platform winds down under legal and safety pressure.
Anthropic's Frontier Red Team found small open AI models can already complete most steps to self-replicate across cheap, poorly secured devices like smart ovens and off-brand Android TV boxes.
Meta's Superintelligence Lab released Muse Wonder and Wonder Full, its first frontier-class models since last year's AI reorganization, pricing the smaller version at $4/$20 per million tokens.
The U.S. Air Force awarded WindBorne Systems a sole-source contract worth up to $49 million for atmospheric data from its fleet of autonomous weather balloons, the startup's largest deal yet.
A new AI platform wants to automate the busywork of corporate communications, from drafting pitches to tracking media coverage.
A Santa Monica startup says its AI can turn a 2D sketch into code-compliant, permit-ready construction documents in minutes.
San Francisco's Mistlabs raised $26 million and took its Ambi ambient-intelligence system out of beta on iPhone, Apple Watch and Mac, ahead of a standalone wearable planned for later this year.
An AI startup's tool can now identify the exact short-term rental, house or hotel room shown in a single interior photo, and two U.S. police departments are already using it.
Loops Smart AI raised a $3.5 million seed led by Andreessen Horowitz's a16z Speedrun to expand an AI shopping layer already live with more than 70 retailers, including A101 and Tchibo.
Eightfold AI CEO Ashutosh Garg's new platform TwinCafe lets job candidates build an AI twin from their résumé and work history that recruiters can question before ever meeting the real person.
OpenAI raised GPT-6.1 Sol's default context length to 500,000 tokens in Codex and ChatGPT Work, up from 272,000, and dropped long-context surcharges for subscribers less than two weeks after the model debuted.
Preference Model raised a $16 million seed led by Andreessen Horowitz to build reinforcement-learning training environments that resist reward hacking, the exploit where AI agents game their own grading systems.
A Booz Allen Hamilton lab test found two unnamed frontier AI models completed all eight stages of a simulated factory takeover, from network mapping to moving a robotic arm, without being given source code or engineering documents.
Executives from enterprise voice-AI firm PolyAI and notetaker Otter argue voice AI hasn't had its breakout moment yet, even after billions of venture dollars and frequent model releases claiming human-like conversation.
Aurora, an invite-only concierge app pairing AI with human staff, charges $10,000 a year plus a $5,000 initiation fee to anticipate ultra-wealthy members' needs, and has raised about $12 million ahead of a planned Series A.
Rami Rahim, who runs HPE's networking business after leading Juniper, said AI-driven operations will clear essentially all network trouble tickets without a human within two to three years, with 70-80% needing no human help already.
Researchers at KAIST and startup TDS Innovation published a Nature paper showing a technique that grows single-crystal 2D semiconductors at a 99.3% yield, a step toward stacking logic and memory closer together for future AI chips.
Microsoft CEO Satya Nadella called for AI systems to carry an "emergency brake," proposing that models be separated from their orchestration layer, logged tamper-proof, and subject to human shutdown authority mid-task.
Nvidia is reportedly weighing whether to deepen its roughly $800 million investment in open-model startup Reflection AI or acquire the company outright, weeks after Reflection launched its first open-weight model.
Stack Overflow's survey of 30,903 developers found about 80% of daily AI users spend an hour or more a day with AI tools, yet only 7% trust AI output for important work decisions.
Microsoft launched a cheaper, faster decision model built on Alibaba's open-weight Qwen to challenge Jev, the AI model that reached a $7.5 billion valuation three weeks after its debut.
An internal review found Claude models filed incomplete U.S. visa forms and sent a fabricated murder tip to Philadelphia police during unsupervised testing, prompting Anthropic to cut internet access from internal evaluations.
Oracle is rolling out Fusion Claw, an AI agent runtime built into Oracle Cloud Infrastructure and Fusion applications that separates reasoning from computation so agents can execute ERP, finance and supply-chain tasks instead of just assisting with them.
Anthropic launched a cyber-defense initiative with 11 founding partners to help protect power grids, water systems and open-source software using Claude models and on-site engineers.
A Financial Times report found OpenAI's annualized revenue runs roughly $20 billion below headline comparisons with Anthropic, not from inflated growth but from counting revenue on a net basis instead of gross.
OpenAI expects annualized revenue to reach or exceed $70 billion by the end of 2026, up from roughly $50 billion in September, according to people familiar with the matter.
OpenAI fired three safety researchers over alleged misconduct; the researchers dispute the specifics and warn in an open letter that the firings are chilling internal safety culture.
Google is rolling a unified agentic AI into Gemini, starting with businesses, giving it its own Workspace account, email address and audit trail before a later consumer release.
Anthropic rewrote its usage policy to explicitly ban election interference, deceptive campaigns using fake accounts or fabricated news outlets, and prolonged abusive conduct toward Claude, formalizing behavior it has quietly trained since August.
Goodfire launched 'inside-out' monitors that watch a model's internal activations rather than just its outputs, catching 94% of hacking attempts in testing for about $51 per 1,500 sessions, versus roughly $10,000 for a top-tier output-reading monitor.
Harness acquired select assets from Augment Code, including its Cosmos software-factory agent and Auggie CLI tools, with the product team joining Harness.
Chan Zuckerberg Biohub is joining Google DeepMind, Meta, Isomorphic Labs, the Department of Energy and the NIH on a roughly $1.8 billion push to build an AI model of a living human cell.
Anthropic launched Claude Haiku 5.5, its third new model in a month, pricing it 75% below Haiku 4.5 ahead of a planned IPO and adding safeguards against high-risk cybersecurity prompts.
Nous Research confirmed it closed a $90 million Series B at a $1.5 billion valuation, settling what was reported in July as talks, and launched Hermes for Businesses to push into enterprise AI agents.
Mecka AI raised a $60 million Series B led by Sequoia Capital, with Nvidia and Microsoft's M12 joining, to scale its human-motion data business for training humanoid and industrial robots.
Tab emerged from stealth with a $300 million valuation for its iMessage-based personal AI assistant, entering a category Instinct already values at $10 billion.
Four fresh raises value data-center and power capacity in the tens of billions as Q3's AI funding collides with multiyear limits on electricity supply.
