Richer than God: Leaked Coatue slides value Anthropic at $2 trillion with $200 billion revenue by 2030 - Newcomer

see, we’re not just throwing cash around on vibes, we got charts and everything

Leak claims Anthropic's Claude source code leaked via a source-map file in its npm registry, prompting investigation, remediation and heightened supply-chain security scrutiny. - X (formerly Twitter)

God mode: Claude Code now has computer use from the CLI on Mac (and desktop app). - Claude

Boris Cherny's favorite Claude Code features - X (formerly Twitter)

Penn economics professor Jesús Fernández-Villaverde used Anthropic's Claude to design, sequence and tutor a personalized master's-level course on sociologist Erving Goffman in about 12 hours and warned such AI threatens university business models. - Business Insider

3 agents in a trenchcoat: Microsoft added Critique and Council multi-model features to Copilot Researcher, combining multiple frontier models including Anthropic and ChatGPT, claiming to outscore either of them independently. - Decrypt

Four AI chatbots, five answers—xAI's Grok, Anthropic's Claude, Google's Gemini and OpenAI's ChatGPT—returned divergent fact-checks of Marco Rubio's claim that U.S. objectives match President Trump's Feb. 28 Iran statement, revealing inconsistency in AI verification. - Fast Company

States including California and Utah are advancing AI regulations, including Gov. Gavin Newsom's executive order requiring safety and privacy guardrails, despite President Trump's December order, with more than 100 state laws and dozens of bills addressing chatbots, deepfakes and security. - The New York Times

Gov. Gavin Newsom issued an executive order requiring safety and privacy guardrails for artificial intelligence companies contracting with the state. - The New York Times

Thomas Rid warns that treating AI as a moral agent misleads and shifts legal liability from humans to firms, prompting calls for clearer contracts, oversight and guardrails that he says will affect Pentagon procurement, regulation, and high‑risk AI deployments. - The Washington Post

Many politicians push to accelerate AI development, but their bullish "unleashing" rhetoric clashes with the sensibilities of many voters. - Financial Times

would slowing down a bit be so bad?

GDPR over GPU compute? Draghi report (Sept 2024) warned the EU faces economic and geopolitical decline, citing AWS that 4 in 10 startups might relocate, electricity costs 2–3× U.S. and natural gas up to 5×, and AI deep adoption has stayed flat. - Yahoo

Grid expectations: David Crane warned that building excess power supply for AI risks stranded assets and urged data centres to bear the costs of generation and risks of grid expansion so capacity matches actual demand, reducing oversupply and unnecessary infrastructure spending. - Financial Times

Chinese open-source AI models are widely adopted by U.S. startups, and can run at roughly one-quarter to one-tenth the cost of U.S. rivals. - Quartz

Check Point disclosed that a ChatGPT code-execution container allowed DNS-based data exfiltration of uploaded files, sending sensitive data to attackers, and OpenAI patched the flaw on Feb 20, 2026 after web controls blocked traffic but DNS egress was overlooked. - The Register

Another malicious npm package compromised by a supply chain attack, highlights need to pin your dependencies so you don’t automatically get latest version deployed. - Snyk

npm install is the new, you’ve won the Wakanda national lottery, click here

Okta builds AI agent verification system with kill switch to prevent rogue actions - The Verge

GitHub removed Copilot's ability to insert "tips" into pull requests after developer backlash, following more than 11,400 instances of automatic promotional edits. - The Register

Major tech firms deploy consumer-facing health LLMs, with Microsoft reporting about 50 million Copilot health questions per day, while independent studies find safety gaps and experts call for controlled studies and independent external validation. - MIT Technology Review

Is an AI algorithm denying your insurance claim? Way more often than you think. - The Palm Beach Post

Law & Order, AI Unit: Crosby, a New York neofirm combining eight AI agents with 30 lawyers, has reviewed about 13,000 contracts for roughly 100 clients, posted ~400% revenue growth since October, and raised $60 million in a Series B at a $400 million valuation. - Forbes

A startup used AI and OpenClaw to stand up a dev team and automate its own developers. - The Wall Street Journal

Options for local and hosted OpenClaw deployment - TechRadar

BTFD-bot: An AI agent automates "buy the dip" trading by executing purchases when prices fall. - The Wall Street Journal

Bots and their owners feel they are being repressed after Wikipedia bans them. - Gizmodo

Back to the future: Cornell instructor Grit Matthias Phelps began requiring, since spring 2023, a once-per-semester proctored in-class assignment on manual typewriters to remove screens, spellcheck and delete keys and curb AI and online-translation use. - AP News

where is the spell check on this thing … correction tape? whoah bro this is a fossil fr

AI umpires add human v. robot drama to major league baseball. - 404 Media

you need glasses clanker?

UFC's Dana White to fans: ‘Shut The F**k Up’ About AI Slop Ads And Just ‘Watch The Fights’ - Kotaku

wrestlers with 6 fingers and four arms are not fake actually

Face/Off - Influencer does face swap of image from US Open to pretend to be at Miami Open. - Where Is The Buzz

Netherlands local election roiled by AI editing of candidate's photo - People

just a little touch-up

New York Times cuts ties with book review freelancer who used AI that seemingly lifted passages from The Guardian. - TheWrap

Follow the latest AI headlines via druce.ai on Bluesky