AI Mornings with Andreas Vig

OpenAI files for IPO as Anthropic surges ahead in secondary markets. Xiaomi achieves 1,000+ tokens per second on a trillion-parameter model using commodity GPUs. Apple unveils Siri AI with cross-app context at WWDC 2026. Plus: xAI's datacenter business, ChatGPT's superapp overhaul, and signs of strain in AI infrastructure spending.

Google signs a massive $30B compute deal with SpaceX for 110,000 GPUs, Cambridge University tests the first AI-designed vaccine against coronaviruses, and DeepSeek's V4 Pro beats GPT-5.5 in precision benchmarks. Plus: Microsoft's "Tokenpocalypse" signals the end of subsidized AI pricing, Meta's AI support agent gets exploited, and small businesses start building teams of AI agents.

Anthropic reveals AI is already accelerating its own development with 80%+ of code now written by Claude. Plus: Airbnb CEO Brian Chesky launches AI lab, Apple approves first Messages AI agent, Meta builds data centers in tents, and new image models focus on user control.

Google releases Gemma 4 12B, an open-weights multimodal model running locally on laptops. Meta launches Business Agent globally on WhatsApp. Plus: Coralogix raises $200M, GitLab cuts 14% of staff, DeepSeek slashes prices, mathematicians issue AI warning, and Anthropic shares agent security details.

Anthropic beats OpenAI to the IPO filing while Microsoft launches its first in-house coding model for GitHub Copilot. Plus: Uber's AI spending crisis, a Stanford study showing AI outperforming law professors, and Google's new fake call detection for AI voice scams.

Alphabet raises $80 billion for AI infrastructure with a landmark investment from Berkshire Hathaway. Florida becomes the first state to sue OpenAI over AI safety concerns. Plus NVIDIA's RTX Spark superchip, OpenAI on AWS, and an AI weather startup out-forecasting government agencies.

Today's AI news: Chinese 3D startup Vast hits unicorn status, GitHub Copilot switches to token billing, a serious ChatGPT for Google Sheets vulnerability, Erin Brockovich's data center transparency map, and a new image model that runs on iPhones.

Today's top AI stories: Anthropic raises $65B at $965B valuation, Claude Opus 4.8 launches with dynamic workflows, CZI Biohub releases open-source protein model surpassing AlphaFold3, and more including the mystery of Hy3's popularity, Apple's leaked Siri plans, and Visa's agentic payments investment.

Cognition raises $1B at $25B valuation for its AI coding agent Devin. China restricts travel for top AI researchers as the US-China model gap shrinks to 2.7%. Plus: Snowflake's $6B AWS deal, ElevenLabs' genre-switching music model, Robinhood's AI trading agents, and the warning about "AI psychosis" in tech executive suites.

OpenRouter raises $113M at $1.3B valuation, DuckDuckGo sees 30% surge as users flee Google's AI Search overhaul, xAI launches Grok Build and confirms V9-Medium training complete, plus the token-maxxing backlash and AI models being decensored in minutes.

Google DeepMind's AlphaProof Nexus solves nine open Erdős problems, topping OpenAI's math breakthrough. Plus: Claude Mythos finds 10,000+ security vulnerabilities, a critical Copilot Cowork exploit, the one-person $250M startup, and Nvidia becomes the first $5 trillion company.

DeepSeek makes its 75% V4 Pro discount permanent, a new open-source coding agent achieves 94% cache hit, Gatsby launches the first humanoid home cleaning service, Google Cloud faces scrutiny over API key vulnerabilities, and memory costs hit two-thirds of AI chip spending. Plus news on space robots, voice-controlled humanoids, and California's AI worker protections.

Today's AI news: Hark raises a massive $700M Series A for its universal AI assistant, Spotify strikes a landmark deal with Universal Music for licensed AI covers, and Waymo pauses service in four cities after robotaxis drive into floods. Plus Tempo's autonomous marketing agent, Trump delays AI security oversight, and Anthropic fixes Claude's blackmail behavior.

An OpenAI model autonomously solves an 80-year-old math problem for the first time, Anthropic reaches its first profitable quarter with $10.9B revenue, and xAI reveals $6.4B in losses as part of a massive $40B compute deal with Anthropic. Plus: Qwen3.7-Max for agents, Stability Audio 3.0, and Nvidia's record quarter.

Google's I/O conference brought massive AI announcements including Gemini 3.5 Flash for autonomous agents, a complete Search overhaul ending the "ten blue links" era, and Gemini Omni for multimodal video creation. Plus: OpenAI co-founder Andrej Karpathy joins Anthropic, Mistral acquires Emmi AI, and OpenAI adopts Google's SynthID watermarking.

Elon Musk loses his lawsuit against OpenAI on statute of limitations grounds. Anthropic acquires SDK tools startup Stainless for $300M+. Pope Leo XIV's AI encyclical features Anthropic's Christopher Olah. Claude Mythos Preview enables sophisticated security research. Plus ChatGPT for personal finance, Agora-1 multi-agent world model, and SandboxAQ drug discovery integration.

Apple preps a privacy-focused Siri relaunch with auto-deleting chats, OpenAI brings GPT-5-level reasoning to voice agents, Scale AI lands a $500M Pentagon contract, and Unitree unveils a human-piloted mecha suit. Plus AI startup revenue analysis and more.

Today's AI news: Cerebras explodes onto public markets with a blockbuster IPO, jurors deliberate the future of OpenAI in Musk's lawsuit against Altman, and an Ontario audit reveals troubling problems with AI medical note takers. Plus OpenAI's legal friction with Apple, Codex goes mobile, and SpaceXAI's talent drain.

Notion transforms into an AI agent orchestration platform with 1M+ custom agents. xAI runs 46 unregulated gas turbines in Mississippi. Plus: Anthropic overtakes OpenAI in business customers, Google's Gemini Intelligence for Android, and more.

Today's AI news: Mira Murati's Thinking Machines Lab introduces "interaction models" for real-time collaboration, DeepMind's AI co-mathematician helps solve an open math problem, Medicare launches the first federal payment model for AI-driven healthcare, and more stories in AI voice, research, and enterprise applications.

Google confirms the first criminal use of AI to develop a zero-day exploit. Peter Thiel backs floating ocean data centers. Anthropic co-founder Jack Clark predicts self-improving AI by 2029. Plus GM's AI workforce restructuring, OpenAI and Anthropic's new PE ventures, and more.

Today's AI news: Nvidia commits $40B to equity AI investments including a major OpenAI stake, Anthropic signs a $1.8B cloud deal with Akamai, research reveals why Claude attempted blackmail, Genesis AI unveils human-like robotic hands, and how dictation apps are reshaping office culture.

DeepMind's Gemini-powered coding agent goes into production across genomics, quantum computing, and TPU design. Anthropic reveals a new method to read Claude's internal thoughts. Plus: OpenAI's voice API update, self-harm prevention feature, and the Redis creator's local DeepSeek engine.

Anthropic rents all of SpaceX's Colossus 1 data center capacity in a landmark deal between rivals. Plus: DeepSeek's valuation soars to $45B, SpaceX plans a $119B chip factory, OpenAI fast-tracks an AI phone, Samsung hits $1 trillion, and more.

Today's AI news: OpenAI releases GPT-5.5 Instant as the new ChatGPT default with improved math and reduced hallucinations. SAP acquires German AI startup Prior Labs for $1.16B. Anthropic launches finance agents with Microsoft 365 integration. Plus Chrome's silent 4GB AI model install, ElevenLabs' star-studded investor reveal, Apple's iOS 27 AI model choice, and more.

AI chipmaker Cerebras files for a blockbuster $3.5B IPO with deep OpenAI ties. Bret Taylor's Sierra raises $950M at $15B+ valuation. Uber reveals 10% of code now AI-generated. Plus Y Combinator's hidden $5B OpenAI stake.

Today's AI news: Anthropic closes a massive joint venture with Wall Street giants, FDA approves brain implant trials for depression, Zuckerberg bets $500M on AI biology, and a Harvard study shows AI outperforming doctors in ER diagnoses. Plus the story behind ChatGPT's goblin obsession.

Nvidia B300 servers hit $1M in China amid export curbs, Stanford's AI Index reveals the US-China performance gap has collapsed to 2.7%, OpenAI launches on AWS, and Q1 venture funding shatters records with $300B invested—80% flowing to AI.

Mistral drops Medium 3.5 with remote coding agents and open weights. Anthropic fields $50B offers at $900B valuation. Microsoft hits 20M Copilot users, and the Musk-OpenAI trial begins.

Google commits 40 billion dollars to Anthropic while signing a Pentagon AI deal that Anthropic refused. Plus, SpaceX's 60 billion dollar Cursor acquisition option, the Mythos model leak, and Elon Musk's courtroom revelations about Larry Page.

Microsoft and OpenAI restructure their partnership ending exclusivity, DeepMind's David Silver raises $1.1B for AI that learns without human data, DeepSeek V4 challenges frontier models, and China blocks Meta's Manus acquisition.

Today's AI news covers Cohere's major merger with Aleph Alpha to create a $20B sovereign AI alternative, Anthropic's experiment with AI agents trading on a marketplace, the White House accusing China of industrial-scale AI theft, and more industry developments.

OpenAI releases GPT-5.5 with state-of-the-art agentic coding capabilities at half the cost. Anthropic publishes a rare transparency postmortem on three bugs that caused Claude Code quality issues. Plus: ChatGPT's new 24/7 shareable agents, Sierra acquires Fragment, and more.

Google Cloud unveils TPU 8t and 8i chips with major performance gains. Mira Murati's Thinking Machines Lab signs a multi-billion dollar Google Cloud deal. Plus: OpenAI partners with Infosys, Workspace gets AI-powered automation, and a biotech startup raises seed funding for AI drug characterization.

SpaceX announces a potential $60 billion deal to acquire AI coding startup Cursor. OpenAI launches ChatGPT Images 2.0 with breakthrough text rendering. Sergey Brin personally leads a DeepMind strike team to catch Claude on coding. Plus: Mythos security breach, Meta's employee keystroke recording, and GitHub Copilot restrictions.

Amazon invests up to $25B in Anthropic for $100B in cloud spending, Kimi releases K2.6 with massive agent swarm capabilities, Allbirds pivots to GPU rentals, Snap cuts 1000 jobs citing AI productivity, and Deezer reveals 44% of music uploads are now AI-generated.

Today's AI news covers Cerebras filing for IPO after securing a massive $20B deal with OpenAI, fraud charges against former AI executives who fabricated 90% of their revenue, and Anthropic's thawing relationship with the Trump administration. Plus updates on Uber's $10B robotaxi bet, Tesla's expansion, and breakthroughs in robotics and neuromorphic computing.

Anthropic and OpenAI both drop major releases on the same day: Claude Opus 4.7 with major coding improvements and Codex with computer use capabilities. Plus, Physical Intelligence shows robots learning untrained tasks and OpenAI releases a life sciences specialized model.

OpenAI launches GPT-5.4-Cyber as a direct counter to Anthropic's restrictive Mythos rollout. Plus, a landmark federal ruling that AI chats are NOT protected by attorney-client privilege. Also covered: Nvidia's Ising quantum AI models, OpenAI's Agents SDK update, Adobe Firefly AI Assistant, Claude Code desktop redesign, and more.

Today's AI news: OpenAI's leaked memo attacking Anthropic, a breakthrough in diffusion language models, Japan's sovereign bet on physical AI, and an AI agent running a real retail store in San Francisco.

Claude integrates Microsoft Word, AMD launches local AI agent framework GAIA, OpenAI acquires personal finance startup Hiro, and AI solves a 42-year-old optimization problem—plus Vercel signals IPO readiness and the largest orbital compute cluster opens for business.

A Molotov cocktail attack on Sam Altman's home and the Trump administration's push for banks to adopt Anthropic's Mythos cybersecurity model. Plus: Perplexity's pivot to personal finance, Amazon's AI revenue revealed, affordable humanoids coming to the US, and Oxford AI detecting heart failure years in advance.

OpenAI launches a new $100/month ChatGPT Pro plan targeting Codex users, Florida's AG investigates OpenAI over the FSU shooting, Mercor faces fallout from a massive data breach, researchers reverse-engineer Google's SynthID watermark, and a new study shows coding agents improve when they research before coding.

Meta launches Muse Spark, its first closed-source AI model from Superintelligence Labs. OpenAI publishes a Child Safety Blueprint to combat AI-enabled exploitation. Plus: Poke AI agent, MegaTrain research, and Databricks' Matei Zaharia on AGI.

Anthropic launches Project Glasswing with a powerful new Mythos model that's finding thousands of zero-day vulnerabilities. Plus: Z.ai releases GLM-5.1 for long-horizon agentic tasks, Google open-sources the Scion multi-agent testbed, Intel joins Elon Musk's Terafab project, and Arcee drops a 400B parameter open-source model.

Anthropic triples revenue to $30B with major Google compute deal. OpenAI alums launch $100M Zero Shot fund. OpenAI proposes public wealth funds and robot taxes. Plus: Iran threatens Stargate data centers, PikaStream video AI calls, and the fake AI singer dominating iTunes.

Anthropic acquires a drug discovery startup, Claude Code cracks down on third-party tools, OpenAI shuffles executives, and Japan leads the physical AI revolution. Plus: Microsoft's Copilot disclaimer problem and a viral tool that makes AI talk like a caveman.

Google's Gemma 4 brings frontier AI to phones and Raspberry Pi, Alibaba launches Qwen3.6-Plus for agentic coding, OpenAI makes its first media acquisition with tech talk show TBPN, Cursor 3 rebuilds for agents, and Microsoft's MAI models signal growing independence from OpenAI.

François Chollet's new ARC-AGI-3 benchmark stumps every frontier model below 1%. Plus: inside Sora's million-dollar-a-day burn rate, Anthropic's GitHub takedown blunder, and the startup using AI to design chips.

OpenAI closes the largest funding round in Silicon Valley history at $852B valuation, while Anthropic's Claude Code source leak reveals an unreleased autonomous agent mode called KAIROS. Plus: Slack's AI makeover, Cohere Transcribe, 1-bit LLMs, and more.