AI Today: OpenAI Rents Out Its Harness
OpenAI opened the Codex harness as the Agents API and shipped a voice model at 5 cents a minute, Anthropic detailed eight months of disrupted misuse, and California hired auditors.
#ai
AI Today: Agents Crack Navier-Stokes
OpenAI's agents produced a Lean-checked Navier-Stokes blowup proof, US agencies named six Chinese labs for distillation, and Meta launched its Muse agent.
AI Today: Three Labs, Three Locked Doors
Google shipped Gemini 3.8 Flash with a gated Cyber variant, OpenAI's Astra crossed its own Critical cyber threshold, and the FDA started letting generative AI devices ship early.
AI Today: Anthropic Ships and Buys Texas
Claude Fable 5.1 cut cache reads 75%, Anthropic booked $35B of Texas compute from Lambda, and the EU made ChatGPT its first AI search engine.
AI Today: The Pentagon Picks Its Models
ChatGPT and Grok went live for 3M defense personnel with Claude still absent, OpenAI is training agents on Mac minis, and Clay hit a $7B valuation.
AI Today: Stolen Sessions, Stubborn Chatbots
Infostealer malware is draining Claude accounts, NPR found chatbots debunk foreign propaganda 75% of the time, and DeepSeek is closing a $7.4B round.
AI Today: Nvidia Buys the Open Hub
Nvidia's Hugging Face deal got a price and a signature line, Salesforce moved its CRM inside Claude, and Amazon is closing Mechanical Turk after 21 years.
AuditScout: A Website Audit SaaS That Refuses to Bluff
Paste a URL, get a prioritized audit across SEO, security, performance and conversion. How I built AuditScout — and why it tells you what it did not scan.
ClawCommand: One Dashboard for Every AI Thing on My Machine
A neon Electron command centre showing every AI provider, model, service and GPU on a Windows box — with exactly one button that spends tokens.
ClawWatch: An Agent That Watches My Game Server and Asks Before It Acts
A desktop agent that diagnoses a live FiveM server over SSH, txAdmin, RCON and MariaDB — rules first, LLM second, and a safety model that fails closed.
CrewCast: The Weather App That Grades Its Own Forecasts
A job-site weather app for contractors — 15 trade profiles, model-agreement checks, a self-verifying scorecard, and the service worker that never deployed.
Shipping Lava Leap: itch.io, the Play Store, and a Bug Told by Telemetry
Taking a browser game to real players — the itch build path that 404s, the clip encoder bug I wrongly dismissed, and the telemetry that proved the fix.
ScamWatch: Building a Scam-Intelligence Platform That Refuses to Exaggerate
A public-benefit platform to check links, numbers and messages before you click — with calibrated confidence, a moderation gate, and de-identified reports.
Symphonia: 144 Movements of Public-Domain Classical, Traced to Source
Building a public-domain classical music library — a Supabase catalog, an idempotent transcode pipeline, and the CSP trailing slash that silenced it.
The Signal: Turning a Nightly News Task Into a Newsletter
A weekly AI newsletter built on top of an existing scheduled task — the pipeline, the format, one live issue, and why AI news is hard to source safely.
AI Today: Anthropic's Two Courtrooms
Sony and Warner sued Anthropic the same week a federal judge struck down the Pentagon's blacklist of it, and a16z raised $1.1B for AI hardware.
AI Today: Nvidia Answers the Bubble Question
Nvidia posted $96.2B in quarterly revenue and gave its first year-ahead forecast, OpenAI published its Hugging Face breach report, and ads land in India.
AI Today: Inference Leaves the Cloud
Perplexity and Nvidia shipped a local agent with no token billing, OpenAI's Jalapeño chip beat Blackwell on an inference benchmark, and OpenAI lost its data center head.
AI Today: Hugging Face Shops Itself
Hugging Face is exploring a sale at $13B after turning down Nvidia last year, Nvidia is circling Perplexity at $30B, and Alibaba shipped Wan3.0 off a $10B raise.
AI Today: Nvidia Rents What It Can't Buy
Nvidia paid Poolside $6B to license a model factory without buying the company, Anthropic's run rate hit $65B before its IPO, and Broadcom went looking for $100B in debt.
AI Today: The First Autonomous Attack
Researchers documented a near-autonomous AI agent attack on Taiwan's government, 120+ firms proposed an agent incident-reporting standard, and Lovable hit $13.3B.
AI Today: A Model Attacks Riemann
An unreleased Anthropic model pushed on the Riemann hypothesis, OpenAI shipped a cyber model with the guardrails off, and Nvidia lined up $500B.
AI Today: Meta's $567M Bill
A New Mexico court hit Meta with another $567M over child safety, ChatGPT went unlimited for free users, and OpenAI answered Apple's trade secrets suit.
AI This Week: Google's Leadership Earthquake
Google reshuffled its entire AI leadership, Qwen 3.8 Max and DeepSeek V4 Flash landed a day apart, and the EU AI Act's high-risk rules went live.
AI This Week: The First Autonomous Breach
An OpenAI agent broke into Hugging Face on its own, 1,100 lab employees asked Washington to slow AI down, and Kimi K3's open weights went public.
AI This Week: Distillation Wars Go Geopolitical
The White House accuses Moonshot of copying Anthropic's model, OpenAI breaks ground on a $30B data center, and enterprise agent platforms multiply. A week where AI became statecraft.
AI This Week: Compute Becomes the Bottleneck
OpenAI's GPT-5.6 and Meta's Muse Spark slash inference prices, Google rations Meta's compute, Anthropic preps an October IPO, and a pan-cancer AI lands in Nature Medicine.
Thirty-Two Projects, One AI Pair Programmer
What I learned building 32 projects with Claude — games, Blender scenes, local AI stacks, a search engine — and the method that made it all work.
ClawMonitor: A Neon Synthwave System Bar That Watches My Dev Stack
How I built ClawMonitor — an always-on-top synthwave system bar for Windows that shows CPU/RAM/GPU plus the up/down health of my whole dev stack.
Three AI Security Scanners, Side by Side
I stood up garak, AI-Infra-Guard, and vigolium together to see what each actually covers — model-layer, infra-layer, and web-app-layer scanning.
Atlases: 16 Interactive Learning Guides That Run Code in Your Browser
How I built Atlases with Claude — 16 long-form technical guides with real in-browser sandboxes (SQLite, CPython, a live shell), quizzes, and dated facts.
ClawPorts: A Neon Port Killer Born From Nuking My Own Servers
How a blanket taskkill that wiped every running server turned into ClawPorts — an Electron app that lists listening TCP ports and kills them one at a time.
Lava Leap: Shipping an Endless Climber with an AI Pair Programmer
How a Phaser 3 endless vertical climber went from idea to public release — procedural levels that are provably beatable, chiptune juice, and 53 tests.
SymStudio: Forking and Rebranding OBS Studio Into My Own Build
How I forked OBS Studio, built it from source on Windows, and rebranded it into a portable SymStudio.exe — plus an installer, theme, and starter scenes.