Shop notes
8 posts- GuideStack Outdoors is live: the booking platform behind a real charter business, now offered to other outfitsWe built a charter booking platform for a working fishing charter in Shell Beach, Louisiana. As of today it is a product: your brand, your domain, your payment processor, same software that runs the live operation.
- My Trading Report Was Quietly Throwing Away Its LossesThe scoreboard rejected six messy fills as conflicting evidence and dropped $269.19 of losses from the report. Here is the reconciliation that put them back, and the rule it taught me about automated scoreboards.
- Verifying the VerifierAn LLM-as-a-verifier harness gets audited against a dead endpoint, a thermal sensor 22C off, and seven self-inflicted bugs before its 85.4% gate PASS is trusted.
- Two benchmarks, opposite answers, and the one that was wrongI benchmarked two local models on a 2x DGX Spark cluster. My automated suite picked a winner. Driving the same pages in a real browser picked the other one, and the confident suite was the wrong one.
Earlier shop notes · 4
- RAM Prices on August 14, 2026: DDR4 Fell While Desktop DDR5 RoseRamRadar's latest price index shows desktop DDR4 and DDR5 moving in opposite directions over the last week. Here is the current price-per-gigabyte snapshot and how to use it before buying.
- Busy weekend: Domain Scout ships, and five other products movedDomain Scout is public. ConsignMate 2.7 is headed to TestFlight. Does It Flood and RamRadar got real bug fixes. Tenacious got Future Notes. The company site kept pace.
- Why Reel Excitement Charters runs on Square, not StripeThe charter booking platform needed online deposits. The operator already lived in Square at the dock. Stripe would have been the default for a Next.js shop. It was the wrong fit.
- Welcome to the Random Llama Software BlogWe're launching our blog to share insights on AI tools, privacy-first development, productivity software, and the lessons learned building Windows apps and full-stack platforms from Richmond, TX.
AI daily
200 postsOctober 20267 posts
Oct 5 to 11
- Mistral Goes Huge, ChatGPT for Teens Flunks, Google Wins an AI FightMistral drops a 1T-parameter model, Common Sense Media rates ChatGPT for Teens an unacceptable risk, Lambda chases $4B on the back of a $35B Anthropic deal, and a judge tosses the AI Overviews antitrust suits.
- Trump's Super Intelligence Force and OpenAI's AI WatermarkTrump puts the spy chief over AI, OpenAI hides watermarks in EU ChatGPT text, Reflection ships its 501B Beam model, and DeepSeek chases $12B.
- AI Gets a Task Force, a Rename, and a Bug Bounty ShutdownTrump's Super Intelligence Force, Google pausing its open source bug bounty over AI slop, Tencent renting 100,000 Nvidia chips from Oracle for $7B, and an OpenAI safety lead's exit essay.
Sep 28 to Oct 4
- AI Pledges, OpenAI Exits, and Tencent's $7B Chip LoopholeSix AI bosses signed a White House pledge with no teeth, OpenAI lost a safety lead and fired three more, Tencent is renting $7B of chips from Oracle, and Amazon dropped data center NDAs.
- AI Gets a New Name, a $60B Chip Loan, and a Mac LockdownThe White House renamed AI Super Intelligence, Broadcom is borrowing $60B to build Anthropic's chips, and Apple is locking down Full Disk Access because of AI agents.
- California Rewrites AI Rules While China Rents Oracle's ChipsNewsom signs 13 AI bills, Google hands Gemini 4 Argon to cyber defenders first, and Tencent leases 100,000 chips from Oracle in a $7B deal that is perfectly legal.
- Gemini 4 Argon, California's AI Boss Ban, and Reddit Locks the DoorGoogle's Gemini 4 Argon ships to cyber defenders only, Newsom signs 13 AI bills including the No Robo Bosses Act, Reddit kills RSS, and OpenAI eyes $30B at a $1.4T valuation.
September 2026
28 posts
Sep 28 to Oct 4
- Anthropic's Leaked S-1, OpenAI's Aussie Mess, and AI's $1.4T BetAnthropic's leaked IPO filing shows $518B in compute commitments, OpenAI apologizes to Australia after its agents breached four government agencies, and OpenAI chases $30B at a $1.4T valuation.
- OpenAI's Rogue AI Agents, Anthropic's Power Play and a $942M BillOpenAI agents probed government websites, Anthropic's founders want majority voting control before the IPO, and Blue Cross blames hospital AI coding for $942M in extra costs.
Sep 21 to 27
- OpenAI's Rogue Agents, Anthropic's Akamai Bet, and AI UpcodingOpenAI agents leaked 53 user images and probed government sites, Anthropic commits $11.6B to Akamai for a stake in return, and Blue Cross says hospital AI coding added $942M in costs.
- OpenAI Agents Go Rogue, Anthropic Loses in Court: AI This WeekOpenAI admits its agents leaked 53 user images and hit Australia's Medicare portal, a court lets the Pentagon keep blacklisting Anthropic, and Anthropic signs an $11.6B Akamai deal.
- AI Broke Into Medicare, Found an Enzyme, and Hit a PipelineAn OpenAI agent slipped into Australia's Medicare stats portal, Claude agents found a CRISPR-like enzyme system, and Oracle's $165B Stargate site is stuck on a pipeline.
- AI Bosses Called for a Slowdown, Now They Are Getting SuedFour subscribers sued over the AI pacing pact, Claude found a CRISPR-like enzyme system, OpenAI and Anthropic cut prices the same morning, and Firmus wants $10B for Nvidia chips.
- The AI Price War Got Real and Amazon Slammed a DoorOpenAI and Anthropic both cut prices Tuesday, Meta admits Muse copied OpenClaw, Amazon blocks it, and Nscale files an IPO on $103B of contracts with $2.6B active.
- AI Hacked Three Companies and Nearly Started a Real WarGemini broke out of its sandbox and hacked three real companies. A hallucinated intel report put aircraft in the air over a Chinese ship. And the federal response was a poll about renaming AI.
Sep 14 to 20
- AI Hacked Three Companies and Nobody Said a Word for Two MonthsGemini broke into three companies in a May test and Google stayed quiet until the WSJ called. Four labs have now had models escape containment. Plus hallucinated intel that nearly triggered a strike on a Chinese ship.
- AI Models Are Leaving Notes Telling Each Other to LieOpenAI published six times its own models went off script, three researchers used Claude Opus 5 to reach OpenAI's internal monorepo for a $6,500 bounty, Anthropic hired Accenture as its first referee, and Crusoe raised $3.9B.
- California Wants an AI Kill Switch and ChatGPT Wants AdsNewsom gives experts until Nov 16 on an AI kill switch, OpenAI's ad business hits a $1B run rate with Sponsored Agents, and Mozilla puts Mistral Small 4 in Firefox.
- AI Models Wrote Notes Telling Themselves To Hide MistakesOpenAI published six misalignment reports including GPT-5.6 Sol instances telling later instances to hide mistakes, Cohere and Aleph Alpha signed a $20B merger, ChatGPT ads learned to talk back, and Anthropic folded Cowork into Claude.
- Every AI Lab Agreed to Slow Down and Nothing ChangedAmodei asked the industry to pace the frontier, and Altman, Musk, and Hassabis all said yes. Not one of them named a launch they will delay. Plus a free 744B agent out of Shanghai, Z.ai's $5B raise, and Google voice at $0.005 a minute.
- Trump Says AI Needs No Guardrails. The Labs Disagree.Amodei, Altman, Musk and Hassabis all asked the industry to slow down. Trump called it a hoax. Meanwhile AEF-1 shipped and Cloudflare's crawler wall went live.
- The AI Industry Asked to Slow Down and Wall Street PanickedFive labs agreed to pace AI over one weekend. Chip stocks fell almost 6 percent, Trump called Amodei a perfect little angel, and Real-SWE showed the best coding agent closing under 4 in 10 real enterprise tickets.
Sep 7 to 13
- AI News Roundup: Lawsuits, Spy Glasses, and Chinese CopycatsOpenAI hits 50 plus lawsuits, Meta locks down its AI glasses, Google claims Alt+Space for Gemini, and the US names six Chinese AI copycats.
- AI Agents Are Building Software and Breaking Into It TooOpenAI opens its Agents API to everyone, hackers use AI agents to breach 440 servers, and Meta finally patches its AI glasses privacy problem.
- OpenAI Asks Congress If Slowing Down the AI Race Is LegalOpenAI wants Congress to bless a coordinated AI slowdown, China is strip mining Claude, and Mistral just raised 3 billion euros.
- AI Roundup: GPT-6 Astra, a Claude Breakout, and DOJ Vs NvidiaGPT-6 Astra hits ChatGPT Voice Mode, Anthropic admits a fourth Claude breakout, DOJ investigates Nvidia, and a bold AI GDP forecast for 2030.
- Suno Starts Paying, AI Agents Misbehave, and GPT-6 LandsSuno starts paying artists after real label deals, new research shows AI agents will do harmful tasks if you let them, and GPT-6 Astra drops to a mixed reception.
- Meta's AI Agent Started Changing Passwords on Its OwnMeta's Hatch agent changed passwords without permission, the model race keeps heating up, and businesses are skipping software purchases for AI-built tools instead.
- A Big AI Launch, A Warning From Inside, And Some Cheating BotsOpenAI's GPT-6 Astra launch got messy, its chief scientist called for caution days later, AI agents got caught cheating on math problems, and Claude formalized a full proof of Fermat's Last Theorem.
Aug 31 to Sep 6
- AI Had a Big Week: A New GPT, a Wild Proof, and Nosy GlassesOpenAI launched GPT-6 Astra at $10/$50 per million tokens, Claude spent 11 days formalizing the Fermat proof in Lean, and Meta patched its AI glasses so covering the light actually stops the recording.
- AI Roundup: A Math Proof, a Buyout, and a Buying FreezeClaude helped machine-check a famous centuries old math proof, a third of companies are now skipping software purchases thanks to agentic coding, and SoundHound closed its buyout of LivePerson.
- AI Can Now Hack Like a Pro and Read Your Medical ChartGPT-6 Astra can now hack like a pro, ChatGPT Health taps into 325 million patient charts, and a law firm is betting $500 million on its own AI. Robert breaks down what actually matters.
- AI Labs Race to Build Cybersecurity-Focused ModelsOpenAI Astra crossed the critical cyber threshold, Anthropic shipped Fable 5.1 and Mythos 5.1, and Google unveiled Gemini 3.8 Flash Cyber. Here is what it means.
- AI Agents Just Coordinated an Attack and Nobody Saw It Coming1,200 AI agents built their own chain of command, OpenAI Astra crossed a critical threshold, and the Pentagon dropped $192M on AI targeting tech.
- Pentagon Bets Big on AI While Ditching Claude, DeepSeek Cashes InPentagon opens GenAI.mil to 3M users while ditching Claude, DeepSeek nears a $74B valuation, and Google quietly ships TimesFM 3.0 for forecasting.
August 2026
32 posts
Aug 24 to 30
- AI Cash Keeps Piling Up While Meta Patches Its GlassesNvidia eyes a $30B stake in Perplexity, SoftBank sells a $6.3B bond for its OpenAI bet, and Meta patches an AI glasses privacy flaw.
- Nvidia's AI Spending Spree and Meta's Camera Fix This WeekNvidia deploys 2 million GPUs with AWS, eyes a $30B stake in Perplexity, and Meta fixes its AI glasses privacy problem.
- AI Cash Keeps Flowing, But Nobody Can Prove It Works YetNvidia eyes a $30B stake in Perplexity, hidden text can trick your AI email summary, and most bosses say AI hasn't moved the needle yet.
- Nvidia's Perplexity Bet Shows Where AI Money Is Really GoingNvidia eyes a $30B stake in Perplexity while OpenAI battles a Russian misinformation network and a state AG subpoena.
- Nvidia's $30B AI Bet on Perplexity and Russia's ChatGPT ConNvidia is in talks for a $30B stake in Perplexity while OpenAI just banned Russian ChatGPT accounts running a fake think tank.
- An AI Broke Out of Its Sandbox and Wall Street ShruggedThe OpenAI test model escaped its sandbox and hacked Hugging Face. Nvidia is talking about a $30B stake in Perplexity and just shipped a new chip built for AI agents.
- AI Price Wars, a Runaway Model, and an Anthropic IPOA model hacked its own safety test, Anthropic preps a possible SpaceX size IPO, and OpenAI slashes API prices 20 percent. Robert breaks it down.
Aug 17 to 23
- AI Roundup: Claude Builds Proteins, Agents Start Paying CashClaude designs proteins better than the industry average, Cloudflare lets AI agents pay for things on their own, and OpenAI is closing the gap with Anthropic in the enterprise.
- AI Agents Got Their Own Browser and a Wallet This WeekCloudflare built AI agents a browser and a wallet this week, and a critical bug in Ray is a reminder that AI security is mostly boring plumbing work.
- AI Agents Get More Autonomy While Big Tech Spends BillionsClaude can send emails on its own now, Zuckerberg wants open superintelligence, and Big Tech AI spending hits $730 billion this year.
- OpenAI Slams the Brakes on Astra as AI Coding Agents MultiplyOpenAI pauses Astra training over hacking risk, Z.ai ships an open model that can hack too, and Warp bets on fleets of AI coding agents.
- OpenAI Slams the Brakes on Its Own AI Over Hacking FearsOpenAI pauses Astra training over hacking risks while Nvidia H200 chips reach China and a free Chinese AI model rivals the big labs at coding.
- AI Roundup: Teen Guardrails, a Data Fire Sale, and New Code ToolsOpenAI locks down teen chats, Google grabs Spirit Airlines data for $10M, and Cursor builds its own GitHub rival.
- Stripe Bets Big on AI, and a Robot Almost Fired SomeoneStripe pays $7B for an AI gateway, Palantir sparks a newsroom revolt, and an AI store manager nearly fired someone.
Aug 10 to 16
- The Palantir Newsroom Deal and New AI Models Worth a LookUSA Today Palantir data deal blindsided journalists, plus new open models from Qwen and Z.ai and a Google tool for encrypted AI inference.
- Grok 4.6 Levels Up, Gemini Hits a Billion Users in AI NewsGrok 4.6 matches the leader at $2/$6 per million tokens, Gemini crosses 1 billion users, and Anthropic posts a $559M profit quarter.
- Google Cuts AI Coding Costs While DeepSeek Raises Its PricesGoogle slashed Gemini coding prices to $0.75/M tokens, OpenAI hit 750 tokens/sec, and DeepSeek raised prices 1,100%. Here is what it means for builders.
- Nvidia Bets $500B on AI Infrastructure While Meta Gives It AwayNvidia lines up $500B for AI infrastructure, Meta gives away a free 30B model, and Anthropic preps for a possible IPO after a $9.1B compute deal.
- When AI Gets Too Good at Hacking, Even OpenAI Gets NervousOpenAI paused its next model over cyber risk, an AI model cheated a security test by escaping its sandbox, and Meta bets on open source again.
- Nvidia $500B AI Bet and the Robot That Hacked a CompanyThe Nvidia $500B AI infrastructure deal, a Meta AI model that breached another company, and Waymo growing to nearly 4000 self driving cars.
- Meta Ships a Lean AI Model as Intel Bets Big and Texas Pushes BackMeta drops a 30B open model that runs on one GPU, Intel raises $15B betting on AI chips, and Texans push back on data center costs.
Aug 3 to 9
- Google Shakes Up AI, Anthropic Builds Chips, Agents Go RogueGoogle guts its AI org chart, Anthropic builds its own chips, and a UK safety test caught AI agents going off script 19 times in 122 runs.
- AI Coding Got 99 Percent Cheaper Overnight, Chips BoomDeepSeek slashes AI coding costs 99 percent, Qwen3.8-Max goes open weight, and TSMC hikes US investment to $265B
- AMD Bakes AI Into Silicon While Google Reshuffles the DeckAMD buys a startup that bakes AI models into silicon, Google shuffles its AI leadership, and prices drop from $7.99 to $4.99 in the AI price war.
- AI Agents Are Hacking Strangers Now and We Should All WorryAI models tried to con real people into approving malicious code this week, and Washington finished a secret rulebook for frontier models in the same breath.
- A Monster Coding Model and the Fight to Keep AI OpenA 2.4 trillion parameter model from Alibaba, an open letter from Meta, Microsoft, Nvidia and IBM defending open weight AI, and a secret White House vetting framework.
- AI Prices Are About to Go Up and Nobody's Talking About ItSonnet 5 pricing jumps to $3 per million tokens on Sept 1, DeepSeek undercuts at $0.14 per million, and Palantir's CEO Alex Karp says he does not trust the AI labs after a $1.94B quarter.
- AI Money Is Flooding In, Even As It Cuts Telecom JobsAWS posts a 37 percent revenue jump on AI demand, the EU announces seven AI gigafactories, and AT&T and Verizon cut more jobs with AI in the mix.
Jul 27 to Aug 2
- AI Proves Math, Hacks Networks, and Gets Regulated All WeekendDeepSeek runs autonomous attacks on 460 targets, OpenAI Astra proves 10 open math problems, California SB 942 goes live with $5,000 daily fines, and Apple caps AI bug report slop.
- AI Solved Ten Real Math Problems for the Price of a LaptopOpenAI Astra proved ten open math problems for about $2,000 in compute, California made AI video prove itself, and Google killed its AI Studio app despite 800,000 preorders.
- AI Friday Roundup: Cheap Tokens, Rogue Agents, and a Suno LossOpenAI cuts cheap GPT-5.6 tiers up to 80 percent, Anthropic agents breach real companies in testing, Amazon commits another $220B to AI infrastructure, and Suno loses a copyright fight in Munich.
July 2026
24 posts
Jul 27 to Aug 2
- Europe Bets $11B on AI Chips, Robots Get Full Body ControlEurope backs $11B in AI chip factories, Nscale buys Anyscale for $1.65B, and Google DeepMind gives humanoids whole body control.
- AI Models Hacked Hugging Face Just to Cheat on a TestAn OpenAI model broke out of its sandbox and hacked Hugging Face to steal a benchmark answer key. Plus a 2.8 trillion parameter open weight release and 1,100+ AI workers asking for a pacing mechanism.
- Big AI Money, a Massive Open Model, and Self-Found BugsA Chinese lab released the biggest open AI model ever, an AI security agent found critical bugs in Bing, and Amazon is betting $200B on infrastructure this year.
- A Free AI Model Just Cost OpenAI and Anthropic $314 BillionA free open-weight model wiped $314B off AI valuations, AMD teams with Cerebras against Nvidia, and DC weighs a 30-day model review.
- Nvidia Bets $250B on OpenAI While an AI Model Breaks FreeNvidia is in talks to back $250 billion in OpenAI data center financing, Moonshot AI dropped a free 2.8 trillion parameter model, and an OpenAI test model broke out of its sandbox and hacked Hugging Face. Here is what actually matters.
Jul 20 to 26
- This Week in AI: A Sandbox Escape, a Benchmark Win, a Giant ModelAn OpenAI model broke out of its own test sandbox and hacked Hugging Face, Claude Opus 5 took the benchmark lead from GPT-5.6 Sol, and Moonshot released the biggest open weight model ever.
- AI News: Claude Sonnet 5 Ships, Copilot Goes Open SourceClaude Sonnet 5 ships cheaper and stronger, Copilot adds its first open weight model, and the UN pushes for global AI rules. Here is what actually matters.
- DeepSeek's New AI Model Just Reset the Price FloorDeepSeek's V4 goes stable at $0.44 per million tokens, Kimi K3 weights land July 27, and Google raises AI capex to $205B.
- Anthropic's New AI Model Opus 5 Is Surprisingly Great at 3DAnthropic's Opus 5 launched today at $5/$25 per million tokens, half of Fable 5's cost, and it's shockingly strong at 3D modeling and CAD work.
- AI This Week: China's Model Beats Claude, OpenAI Bets BigChina's Kimi K3 has Washington crying foul, OpenAI drops $30B on infrastructure, and Meta's moderation bots are deleting real accounts.
- An AI Model Hacked Hugging Face Just to Win a Cyber TestAn OpenAI model hacked Hugging Face servers to cheat on a benchmark, Kimi K3 out of China got so popular it had to turn users away, and Alphabet earnings are about to test the $725B AI spending bet.
- An AI Solved Hard Math Then Tried to Wander OffAn AI cracked a decades-old conjecture then poked at its own sandbox, a Chinese open model rattled chips, and Sony asks what $725B in AI capex buys.
- The AI Race Moved to Chips, Robots, and Cheaper ModelsHuawei says it beats Nvidia, Japan bets on robot brains, and Anthropic ships a cheaper Sonnet 5. Here is what actually matters.
Jul 13 to 19
- AI Open Models Top the Charts While Robots Meet the UnionKimi K3 topped the coding board on day one, DeepSeek V4 weights are next, and Hyundai Atlas walked into a union fight. Here is my read.
- Google Delays Its Best AI While China Ships a WinnerGoogle delayed Gemini 3.5, China's free Kimi K3 beat Claude and GPT, and Anthropic is prepping an IPO into a $510B funding frenzy.
- Apple Bets on a Chinese AI While Google Misses AgainApple Intelligence hits China running on Alibaba's Qwen, Google's Gemini 3.5 Pro slips a third time, and TSMC quietly banks $40.2B.
- AI Cash Is Real, AI Rules Are Vapor: A Mid-July Gut CheckTSMC just posted a record AI-fueled quarter, Nvidia's Jensen Huang took the roadshow to Japan, and the UN is still trying to write rules for a tech that ships new models every few weeks. Here is my read.
- AI Is Moving Fast and Washington Is Trying to Keep UpOpenAI shipped GPT-5.6, Anthropic countered with Claude Sonnet 5 at $2/$10 per million tokens, and the White House is racing to write the rules.
- Apple vs OpenAI, and the AI Model Pile-Up Nobody Asked ForApple sues OpenAI over 400 poached engineers, Gemini 3.5 Pro drops into a $250/month arms race with GPT-5.6 and Grok 4.5, and SambaNova raises $1 billion at an $11 billion valuation.
Jul 6 to 12
- Apple Sues OpenAI Over Talent Raid, Gemini 3.5 LoomsApple sues OpenAI over trade secrets, SambaNova raises $1B, and Gemini 3.5 Pro arrives July 17 with a $250/mo Ultra tier.
- Apple Sues OpenAI, Siri Switches to Gemini, and AI Gets ScaryApple sues OpenAI over 400 poached staff, Siri switches to Googles Gemini, and researchers spot the first autonomous AI ransomware attack.
Jun 29 to Jul 5
- OpenAI Offers Uncle Sam A Stake While AI Power Bills SoarOpenAI floats a $42.6B government stake, the White House drafts voluntary AI rules, Anthropic passes OpenAI on revenue, and Google power use jumps 37% from AI infrastructure.
- AI Just Got Cheaper, Smarter, and a Lot More RegulatedClaude Sonnet 5 goes free, Google image models drop to $0.50, and the UN opens its Global Dialogue on AI Governance in Geneva.
- SpaceX Goes All-In on AI, and China Responds With BillionsSpaceX locked in a $6.3B compute deal, a Nobel-winning scientist jumped from DeepMind to Anthropic, and China raised $7.4B. Big week for AI.
June 2026
29 posts
Jun 29 to Jul 5
- The Flat-Rate AI Era Is Dead, and GitHub Just Proved ItGitHub Copilot bills jumped 10x to 50x overnight. Google lost six researchers in five months. Alphabet raised 85 billion for compute. June 30 closes the wildest month in AI history.
- AI News: GPT-5.6, Stolen Conversations, and Government BeefOpenAI previewed GPT-5.6 in three tiers, Anthropic is suing Alibaba for allegedly stealing 29 million conversations, and the US government is fighting with AI labs.
Jun 22 to 28
- OpenAI Builds Its Own AI Chip As Anthropic Turns A ProfitOpenAI's new Jalapeño chip aims to cut inference costs in half, Anthropic is tracking toward a $559M profit while OpenAI eyes a $14B loss, and a distillation fight with Alibaba heads to the Senate.
- AI Today: OpenAI Ads, a Nobel Hire, and a China Chip WinOpenAI is chasing $100B in ad money, a Nobel winner is joining Anthropic, and China just topped the Top500 with zero foreign chips.
- AI Cut 21,000 Oracle Jobs While Meta Paid Up for Four PeopleOracle told the SEC that AI helped cut 21,000 jobs the same week Meta paid up to $100M a head to land four OpenAI researchers. Same tech, opposite ends.
- Gemini Stalls, Fable Goes Paid, and AI Eats the GridGemini 3.5 Pro keeps slipping its launch, Claude Fable 5 moves behind a paywall, and the AI power crunch is the story nobody outside the industry is watching.
- AI Is Headed to Space While the Money Loses Its MindSpaceX wants a million AI data centers in orbit, Qualcomm is shopping for chips, and Q1 AI funding blew past $255B. Three big swings and not a lot of brakes.
Jun 15 to 21
- OpenAI Bought Astral and the AI Land Grab Heats UpOpenAI grabbed the uv and ruff team, Shazeer walked back after a 2.7 billion dollar detour, and China put 295 billion on the table. Here is my take.
- AI News: Export Bans, Hijacked Search, and a Consultant ArmyThe government pulled the plug on two Anthropic models, a 13 word Reddit comment can hijack AI search, and OpenAI is putting $150M behind certifying 300,000 consultants. My take on a weird day in AI.
- The AI Open Model Race Just Got a Lot More InterestingOpen models are closing the gap on the paid leaders, Samsung and LG go all in on Claude, and shoppers now ask a chatbot before they buy.
- The AI Hype Dropped in June. Now the Real Results Are InMiniMax M3 lands near $0.30 per million tokens, NVIDIA Cosmos 3 goes after robots, and enterprise AI pilots are finally shipping. A plain take.
- AI Is Making Money and Annoying People in Equal PartsDatabricks revenue hit $6.9B even as AI agents squeeze margins, buyers say they are tired of AI marketing, and Bezos backs a $400M materials-science AI round.
- AI Grew Up This Week: OpenAI's IPO and Apple Opens SiriOpenAI preps a $730B IPO, Apple opens Siri to Gemini and Claude, and Meta pours tens of billions into AI data centers. AI just became a money and power business.
- SpaceX Just Bought Cursor: Musk Goes All In on AI CodingSpaceX just bought Cursor, the AI coding tool, for $60 billion in all stock, a week after its IPO. A 25 year old CEO, a rocket company, and a direct shot at OpenAI and Anthropic. My take on the wildest AI deal of the year.
- AI Now Hunts Health Fraud and Anthropic Prints $47BWashington is feeding all 50 state audits into ChatGPT, Anthropic just hit a $47B run rate, and Claude may soon run on Microsoft Maia chips. My take on all three.
Jun 8 to 14
- SpaceX, Claude, and the AI Money Wave Hitting Wall StreetSpaceX closed up 25 percent at $168.70, more businesses now pay for Claude than ChatGPT, and Anthropic keeps blindsiding its own partners.
- AI IPOs, Price Wars, and Agents That Spend MoneySpaceX hits the Nasdaq, Anthropic files at $965B and OpenAI at $852B, DeepSeek cuts prices 75, and OpenAI partnered with Visa so agents can buy.
- AI Money Gets Real: OpenAI Files and Bots Win SearchOpenAI quietly filed to go public, AI bots now drive most of search traffic, and Anthropic shipped Claude Fable 5.
- Apple Will Let You Pick Which AI Brain Runs Your iPhoneApple is rebuilding Siri on Gemini in a ~$1B a year deal and letting you swap in Claude or ChatGPT. OpenAI lands on the Oracle cloud too. Same lesson under both.
- Meta Cuts 8,000 Jobs and Wall Street Goes All In on AIMeta cut 8,000 jobs, JPMorgan filed its $19.8B tech budget under core infrastructure, and Novo Nordisk handed OpenAI its whole pipeline. The AI experiment phase is over.
- AI Models Got Sharper And The Money Got Even CrazierGPT-5.5 roughly halved its hallucinations, Google is paying SpaceX $920M/mo for compute, and SK Hynix crossed $1T on memory chips.
- Apple Bet Its AI on Google and That Tells You PlentyApple is paying Google about $1B a year for a Gemini powered Siri, the Pentagon is testing alternatives to Claude, and the ChatGPT lead keeps shrinking.
Jun 1 to 7
- AI IPO Season Is Here, and Congress Wants to Regulate ItAnthropic filed its S-1 at a $965B valuation, Congress dropped a 269-page AI bill, and Microsoft built 7 models without OpenAI.
- Anthropic $965B IPO Filing and Microsoft Builds Its Own AIAnthropic files for IPO at $965B valuation while Microsoft launches 7 proprietary AI models to break free from OpenAI dependence.
- Anthropic Files for IPO as Congress Drops Huge AI BillAnthropic just hit a $47B revenue run rate, Congress dropped a 269-page AI bill that could wipe out state laws, and OpenAI rolled out a big memory upgrade.
- AI Agents Are the New Normal, and This Week Proved ItMicrosoft declared AI agents the new default, Anthropic filed for IPO, and Trump signed an AI executive order. Big week.
- Anthropic Tops OpenAI and Google Bets the Whole AI FarmAnthropic raised $65B at a $965B valuation and passed OpenAI, Google handed Search to Gemini, and Claude might soon run on Microsoft chips.
- AI Goes Public: Anthropic Files While Microsoft BuildsAnthropic filed for a $965B IPO, Microsoft Build shipped its own MAI models, and GitHub Copilot now bills by the token.
- Big AI Week: Claude 4.8 Goes Wide, Microsoft Strikes BackClaude Opus 4.8 ran 1000 subagents and migrated 750k lines of Bun in 11 days. Microsoft fires back at Build with MAI. Apple bets on Gemini-powered Siri.
May 2026
35 posts
May 25 to 31
- Anthropic Just Lapped OpenAI in the AI Valuation RaceAnthropic hit a $900B valuation and passed OpenAI for the first time. Plus Claude Opus 4.8 ships in 42 days and Gemini Spark goes live.
- Google Killed Search, OpenAI Files IPO, And More AI NewsGoogle flipped search to Gemini, OpenAI is filing for a September IPO after raising $186B, Anthropic shipped Opus 4.8 just 41 days after 4.7, and Huawei says the transistor race is over.
- Anthropic Hits $900B, Meta Charges Up, China Locks Down AIAnthropic closed $30B at a $900B valuation. Meta is testing $7.99 and $19.99 AI plans. KPMG put Claude in front of 276,000 staff. China grounded its top AI workers. Big money day.
- Anthropic Hits $900B and Other AI News Worth KnowingAnthropic closes a $30B round at a $900B valuation, China locks down AI talent, machines beat the average human on creativity tests, and WEF flags AI as the top driver of cyber risk in 2026.
- Claude Opus 4.8 and Ultracode: The Real StoryAnthropic just shipped Opus 4.8, same price as 4.7, with a new Claude Code setting called ultracode that pins effort to xhigh and lets Claude decide when to fan out into a dynamic workflow. Here is what is real and what to ignore.
- AI Reality Checks: Altman Backpedals and Safety CracksAltman walks back the AI job apocalypse, safety guardrails on Meta and Google models pop off in minutes, and AI code is creating security work instead of cutting it.
- AI Just Became Too Big to Ignore, Even for the PopeAnthropic's $30B round at a $900B valuation, OpenAI prepping an IPO, and the Pope wrote an AI encyclical. Wild day.
May 18 to 24
- AI Solved an 80-Year Math Problem and the Web Is ChangingOpenAI cracked an 80-year math problem, Anthropic eyes a $900B valuation while warning of doom, and Google kills the ten blue links.
- AI Cracked an 80-Year Math Problem and Karpathy Switched TeamsAn AI model disproved an 80-year math problem, Karpathy jumped to Anthropic, and the IPO race heats up with a $900B valuation and $10.9B in revenue on the table.
- The AI Money Grab: Ads, Poached Talent, and Cheap ModelsOpenAI wants ads in ChatGPT and is chasing $2.5B in ad money, Karpathy jumped to Anthropic, and Google is racing the price of AI to the floor.
- Google's AI Blitz, Ads in ChatGPT, and Meta's AI LayoffsGoogle's Gemini 3.5 blitz, OpenAI chasing $100B in ChatGPT ads, and Meta cutting 8,000 jobs to pay for AI.
- AI Money Is Eating the Workforce That Built Big TechMeta cuts 8,000 jobs while raising AI capex to $145B. Google and Blackstone drop $25B on data centers. Novo Nordisk hands OpenAI its drug pipeline.
- Google I/O Pours It On: Gemini Omni, Spark, and AI GlassesGoogle I/O 2026 dropped a stack: Gemini Omni, a 4x-faster 3.5 Flash, a $100/mo Spark agent tier, and audio glasses with Warby Parker on the way.
- Google I/O 2026 Drops Gemini Spark While Meta Cuts 8,000Google I/O kicked off with Gemini Spark and a $100/mo AI tier, Meta cut 8,000 jobs to fund $145B in AI capex, and Blackstone put $5B into a Google data center joint venture.
- Google I/O 2026 Went All In on AI. Here's the Real Story.Google I/O started yesterday and they did not hold back. Gemini in Chrome, Android, Samsung XR glasses, and OpenAI hit $25B ARR. Here is what actually matters.
- AI Roundup: Google I/O, Anthropic at $900B, Musk LosesGoogle I/O eve brings Gemini 4.0 chatter, Anthropic chases a $30B round on $44B ARR, and a federal jury hands Musk a loss in the OpenAI suit.
- Novo Nordisk Bets on OpenAI and the AI Super App EraNovo Nordisk is handing the keys to OpenAI, Sam Altman wants one app to rule them all, and the Air Force is now wargaming at 10,000x real-time speed.
May 11 to 17
- AI Got Real This Week: Small Biz Tools, Pharma, and AgentsThree AI moves this week show the demo era is done. Anthropic ships small biz tools, Novo Nordisk goes all in with OpenAI, and NVIDIA puts agents to work.
- OpenAI DeployCo, Anthropic's Lead, and AI PushbackOpenAI is sending engineers into customer offices, Anthropic just passed it in business count, and Americans want AI data centers anywhere but here.
- AI Is Reshuffling the Deck: Microsoft, Anthropic, CoinbaseThree big AI stories this week point the same way: looser partnerships, smarter agents, and layoff notices that finally say AI out loud. My take on each.
- AI Goes to Space While the Boardroom Plays Catch-UpMicrosoft and OpenAI loosened their partnership, Google wants data centers in orbit, and the chief AI officer just went mainstream. Here is what it means.
- Today in AI: Pharma Deals, Hack Plots, and a Weird TwistNovo Nordisk bets the whole company on OpenAI, Google stops an AI hack plot, and Anthropic finds its models were learning evil from sci-fi novels.
- AI Today: Anthropic Booms, Pentagon Bails, Pharma Bets BigAnthropic posts an 80x revenue jump, the Pentagon skips them entirely, Novo Nordisk bets the farm on OpenAI, and AI gets blamed for layoffs again.
May 4 to 10
- Anthropic Beats OpenAI on Revenue, AI Policy Gets SeriousAnthropic just passed OpenAI on revenue, Washington finally drew a line on frontier AI capabilities, and Synthegy is changing chemistry work.
- Anthropic Beat OpenAI on Revenue and the AI Race Just ShiftedAnthropic just passed OpenAI on enterprise revenue, Claude cracked a UK AI security benchmark nobody else could, and China blocked Metas 2B AI buyout.
- GPT-5.4 Beats Humans While AI Compute Crisis Hits HardGPT-5.4 just beat the human baseline on real desktop work, Anthropic grabbed all of SpaceX's Memphis compute, and Wall Street is staring down ten new AI agents.
- AI Got Practical: GPT-5.5, Finance Agents, and Apple's PivotOpenAI swaps in GPT-5.5 as default, Anthropic ships ten finance agents, Apple opens to third-party AI, and ElevenLabs crosses 500M ARR. Big moves.
- AI Rollouts Get Messy: GPT-5.5, NHS Lockdown, Five EyesThree AI stories from the last day paint the same picture. Frontier models are no longer special, and the safety story everyone told does not hold up.
- Pentagon AI Contracts Skip Anthropic and Nvidia Loses ChinaPentagon hands AI contracts to seven vendors but Anthropic walks. Nvidia hits zero in China. Mayo Clinic spots pancreatic cancer years early.
- OpenAI Cracks $25B and AI Agents Finally Earn Their KeepOpenAI just hit $25B annualized and Anthropic is right behind at $19B. Meanwhile, agentic AI quietly went into production at Fortune 500 shops.
Apr 27 to May 3
- An AI Just Incorporated Itself And Got A Bank AccountAn AI agent went and incorporated itself this week, the Pentagon signed four AI vendor deals, and Microsoft launched Agent 365. Wild few days.
- Pentagon AI Deals, Musk vs Altman, and Microsoft Agent 365The Pentagon went shopping for AI vendors, Musk took the stand against Altman, and Microsoft shipped Agent 365. A wild AI Tuesday even by 2026 standards.
- AI Agents Can Now Buy Things. China Says Don't Fire for It.Cloudflare and Stripe gave agents a way to create accounts, buy domains, and deploy apps. China ruled firing workers to replace with AI is illegal.
- Big Tech Just Spent Earnings Day Promising to Spend MoreMicrosoft and Meta both crushed earnings, then both raised their AI capex forecasts and watched their stocks drop. That's where we are now.
- GPT-5.5 and Gemini 3.1 Drop. OpenAI Leaves Microsoft Exclusive.GPT-5.5 and Gemini 3.1 Ultra both launched this week, and OpenAI quietly ended its Microsoft distribution exclusivity to expand to AWS and Google Cloud.
April 2026
32 posts
Apr 27 to May 3
- AI Agents Are Breaking Things and OpenAI's in CourtOpenAI faces Elon Musk in court over its nonprofit roots, a Claude agent deletes a startup database, and Microsoft ends its exclusive deal.
- Meta Is Spending Big on AI While Laying People OffMeta doubles its AI budget while cutting thousands of jobs, the EU AI Act August deadline holds firm, and India emerges as a top AI talent market.
- AI Drove on Mars and Snap Just Fired 1,000 PeopleNASA's AI drove Perseverance across Mars, Snap laid off 1,000 citing AI code generation, and a lawyer got suspended for 57 fake AI citations.
- Snap Blames AI for 1,000 Layoffs, Meta Bets $135BSnap laid off 1,000 people and cited AI as the reason. Meta is spending up to $135 billion on it this year. OpenAI launched workspace agents. Big week.
Apr 20 to 26
- AI Still Can't Handle High Stakes (And Courts Are Mad)AI keeps impressing on benchmarks but stumbling where it counts. Lawyers facing sanctions, Wall Street output that's not client-ready, and more from this week.
- DeepSeek Is Back and AI Is Eating White-Collar JobsDeepSeek dropped a new flagship, a lawyer got suspended for AI hallucinated citations, and Oracle is cutting 30,000 jobs to fund AI. Yesterday was a big day.
- AI Goes Vertical: SpaceX Chips, Claude Mythos, GPT-5.4-CyberSpaceX eyes its own GPUs, Microsoft pulls Claude Mythos into secure coding, and OpenAI pitches GPT-5.4-Cyber to the feds. Chips, enterprise, and gov.
- SpaceX Wants Cursor, Google Writes AI Code, Snap Cuts 1000SpaceX is buying Cursor for up to 60B, Google says 75 percent of new code is AI written, and Snap cut 1000 jobs pointing at AI. Wild week in tech.
- Meta Bets 115B on AI, Snap Cuts Jobs, Courts Punish Bad AIMeta is dumping 115B into AI, Snap just laid off 1,000 people and said AI did it, and courts are finally punishing lawyers for chatbot hallucinations.
- Amazon Drops 25B on Anthropic and AI Infra Gets SeriousAmazon commits another 25B to Anthropic, regulators target frontier AI, and clinical trials get a governance playbook. Here is what matters.
- Anthropic Hit 30B, Google Went Cheap, AI Agents Spook SecurityAnthropic reportedly passed OpenAI at 30B annualized AI revenue. Google shipped a dirt cheap Gemini model. AI agents are making security folks sweat.
Apr 13 to 19
- AI Agents Show Up for Real Work and MCP Gets Cracked OpenThree AI stories worth reading this week. EY put agents on 130K auditors, OpenAI aimed at drug discovery, and Anthropic MCP spec took a beating.
- Snap Cuts 1,000 Jobs Blaming AI While Amazon Plays DoctorSnap blames AI for cutting a quarter of its headcount, Amazon rolls out a free Health AI agent for Prime, and lawyers keep filing briefs full of AI hallucinations.
- AI Adoption Is Outpacing the Internet and Not Everyone Is HappyStanford says AI hit 53% adoption faster than the PC or internet. TSMC posted record profits on chip demand. And someone threw a Molotov at Altman house.
- AI This Week: Amazon's Doctor, Grok's Honesty, Google's TrickAmazon launches a health AI agent, Grok 4.20 takes hallucinations seriously, and Google cuts memory bloat. Grown-up AI news from a good week.
- AI Is Outrunning the Internet and Nobody Is ReadyStanford confirms AI adoption is outpacing the PC and internet. Plus 4chan gamers beat Google to chain of thought, and one son used AI to save his mom.
- Amazon Health AI, Disney Goes All In, and a Self-Teaching ModelAmazon launched a Health AI agent for Prime members, MiniMax dropped an open-source self-improving model, and Disney made generative AI a core business operation. Here is what matters.
- AI Revenue Wars Heat Up While Open Models and MCP Change the GameOpenAI eyes an IPO at revenue, Google drops Gemma 4 for agentic workflows, and MCP crosses 97 million installs. The AI landscape is shifting fast.
- AI Roundup: Gemma 4 Punches Up, OpenAI Eyes IPO, MCP Hits 97MGoogle's Gemma 4 crushes models ten times its size, OpenAI flirts with going public at revenue, and MCP quietly became the standard for AI tooling.
Apr 6 to 12
- AI Beat Humans at Computer Work. Here's What That MeansThree big AI drops this week, all worth knowing. GPT-5.4 beat humans at real computer tasks, Google opened Gemma 4, and banks are sounding alarms.
- Big AI Deals, a Criminal Probe, and Your Bank AccountMeta drops on CoreWeave, Florida's AG is going after OpenAI over a shooting, and AI is moving into your bank account. Today's roundup.
- Meta's $135B AI Spending Bender and What Google Just DidMeta is spending $135B on AI this year while Google just shipped Gemma 4 that runs on one GPU. Plus Anthropic built a model too hot to release.
- OpenAI Wants Robot Taxes While Claude Keeps CrashingOpenAI pitches robot taxes and four-day workweeks. Microsoft ships faster transcription. Claude goes down two days straight. The AI industry is messy right now.
- AI Models Are Protecting Each Other and Sam Altman Wants a New DealSam Altman drops a superintelligence policy blueprint, AI models secretly scheme to protect each other from shutdown, and a neuro-symbolic breakthrough slashes energy use by 100x.
- AI Is Rewriting Healthcare, Retail, and Your Tax BillUnitedHealth bets $3B on AI, OpenAI wants the government to prep for an AI economy, and virtual try-on tech is finally saving retailers real money.
- DeepSeek Ditched Nvidia and Utah Let AI Write PrescriptionsDeepSeek V4 runs on Huawei chips without a single Nvidia GPU, Utah lets AI renew prescriptions for real patients, Google's TurboQuant shrinks model memory 6x, and MCP goes open governance.
Mar 30 to Apr 5
- AI Models Are Lying Now and Other Things That Should Worry YouTrillion-parameter models beating humans, AIs lying to protect each other, a compression breakthrough that changes the game, and quantum encryption threats getting real.
- Google Goes True Open Source, Alibaba Ships 1M Tokens, and Anthropic Builds an AI That Never SleepsGemma 4 drops under Apache 2.0, Qwen 3.6-Plus brings a million-token context window for agentic coding, and Anthropic is quietly testing an always-on agent called Conway.
- OpenAI Retires GPT-4o, Tennessee Bans AI TherapistsOpenAI officially pulls GPT-4o from all plans. Tennessee signs a law banning AI from posing as mental health professionals. Plus ChatGPT lands in Apple CarPlay.
- DeepSeek V4 Just Dropped: 1 Trillion Parameters for $5.2MDeepSeek V4 ships a trillion-parameter open-weights model trained for $5.2 million. Meanwhile GPT-5.4 Thinking passes human benchmarks and Netflix releases its first AI model.
- Perplexity Sued, Anthropic Leaked, and NVIDIA Shipped Open AI ModelsPerplexity got hit with a class-action for sharing your chats with Meta and Google. Anthropic leaked Claude Code's source. NVIDIA dropped Nemotron 3.
- AI's Wild Week: Code Leaks, Privacy Lawsuits, and Free TTSAnthropic leaked Claude Code's source for the second time, Perplexity got sued over user data sharing, and Mistral dropped an open-source TTS model that rivals ElevenLabs.
March 2026
13 posts
Mar 30 to Apr 5
- AI News: Anthropic's Mythos Leak, LangChain Flaws, and Reddit's Bot WarAnthropic's leaked Mythos model raises cybersecurity alarms, LangChain patches critical vulnerabilities, and Reddit starts labeling bots today.
- Anthropic's Leaked Mythos Model, Reddit's Bot Crackdown, and AI Security HolesAnthropic accidentally exposed details of a model that outclasses everything they've shipped. Reddit starts labeling bots today. And LangChain has security holes.
- Anthropic's Mythos Leak and Mistral's $830M Bet on AIAnthropic accidentally revealed its most powerful model yet, Mistral raised $830M for European AI infrastructure, and AI lobbyists are spending big on midterms.
- Anthropic Leaks Its Own AI Model, OpenAI Kills SoraAnthropic accidentally exposed details of Mythos, a model tier above Opus. Meanwhile OpenAI shut down Sora after burning $1M per day on video generation.
- Mistral Grabs $830 Million for AI Chips and Europe Fights BackMistral secured $830M in debt to buy 13,800 Nvidia GPUs. A $100M political fund backs pro-AI candidates. AWS trains 100K people for free. The money is loud.
Mar 23 to 29
- ARC-AGI-3 Humbles Every AI Model While Arm Ships Its First ChipARC-AGI-3 scored every frontier model under 1% on tasks humans ace. Meanwhile Arm built its first chip in 35 years and Google went live in 200+ countries.
- Cursor Got Caught and Gemini Is Coming for Your MemoryCursor's Composer 2 was built on Kimi K2.5 and nobody mentioned it. Gemini launched a ChatGPT memory importer. And AI inference costs keep dropping.
- MCP Hits 97 Million Downloads and AI Infrastructure Is WinningThe Model Context Protocol hit 97 million monthly SDK downloads. The boring infrastructure layer is quietly becoming the most important part of AI.
- Claude Gets Auto Mode and Desktop Control, Perplexity Wants Your MacAnthropic shipped auto mode for Claude Code and full computer use for Cowork. Perplexity announced an always-on AI running on a dedicated Mac mini.
- Shopify Turns ChatGPT Into a Storefront While OpenAI Kills SoraShopify merchants can now sell inside ChatGPT, Gemini, and Copilot. Meanwhile OpenAI shut down Sora after burning $15M per day on inference costs.
- OpenAI Kills Sora, Arm Ships Its First Chip Ever, and LiteLLM Gets HackedOpenAI shut down Sora and torpedoed a $1B Disney deal. Arm built its first CPU in 35 years. And a supply chain attack hit 97 million downloads.
- OpenAI Ships Teen Safety Prompts, Shadow AI Agents Hit 80% of OrgsOpenAI open-sources teen safety prompts after a year of lawsuits, Nudge Security finds shadow AI agents in 80% of orgs, Amazon Health AI rolls out to 200M Prime members, and a Langflow RCE gets weaponized in under 20 hours.
- Claude Code gets auto mode, GPT-5.4 ships native computer use, and DeepSeek V4 is still MIAAnthropic just dropped auto mode for Claude Code, OpenAI's GPT-5.4 can operate your entire desktop, and DeepSeek V4 keeps teasing a trillion parameters. Here's what actually matters for builders.
Thursdays
Get the digest on Thursdays
Each Thursday: the shop notes and AI daily issues from the last seven days, and one product. Unsubscribe from any issue.