Quid Pro Claude: Anthropic CEO Dario Amodei heads to the Commerce Department Monday; tech leaders signed an open letter urging officials to lift the export controls - CNBC

The White House Is Ratcheting Up Its War Against Anthropic: This is how America loses the AI race. - The Atlantic

How the Anthropic saga could threaten American AI dominance - Axios

Washington’s suspension just strengthened Beijing’s pitch for its own competing AI models - Financial Times

The White House gave Anthropic a 90-minute ultimatum over a jailbreak that cybersecurity experts say was the model working as intended. - The Verge

How Anthropic lost the White House's trust - The Washington Post

by not donating $25m like Brockman - apparently now "the White House's trust" is a prerequisite to shipping products, even to vetted customers. what do y’all not get about pay-to-ship

Silicon Valley VCs who railed against Biden AI safety rules are silent about Trump's - Techdirt

“Constitutional AI” in crisis: Anthropic's origin story is rooted in not trusting others with safety - Stratechery by Ben Thompson

MAGA doesn’t like it any more than the real Constitution

y’all are so into safety, safest thing is don’t ship

Dario does lecture them like they are idiots and is mostly right which is infuriating and unforgivable … Dems might just nod when e.g. Andreessen lectures them but MAGAs are made different

Your AI wants to card you: Anthropic updates TOS to allow it to check your ID - theregister

Anthropic paused billing changes that would have disqualified 'claude -p' and Agents SDK usage under Max subscriptions and forced API token billing. - 4sysops

Maximally misleading: Anthropic faces class action lawsuit alleging Claude Max subscribers were misled about usage limits and token costs. - 4sysops

one one hand Claude Max is a mystery box of chocolates of varying usage quota of varying quality, you never know what you're gonna get, on the other hand it's heavily subsidized compared to the API so 🤷‍♂️ - Reddit

Meta CTO Bosworth calls the company's AI reorg "atrocious," promises fixes including smaller management spans and better perks - WIRED

Meta launches AI Mode in Facebook search, generating conversational answers from public posts and Reels with no opt-out clarity yet - TNW | Apps

AWE 2026 spotlights the smart glasses boom: 2.25 million units shipped in Q1, up 167%, with Meta holding 69% market share - CNET

Meta must face lawsuit alleging it torrented 2,300 copyrighted adult films for AI training - Mashable

gold mine that drove into a clown car? (Zuck used to call Twitter the clown car that drove into a gold mine)

OpenAI's 2025 net loss hit $38.5 billion on $13 billion revenue as the company eyes a $1 trillion IPO. - Ed Zitron's Where's Your Ed At

Judge dismisses xAI's trade secret lawsuit against OpenAI, Musk's second legal loss in a month - Reuters

ChatGPT's market share fell below 50% for the first time in May 2026, while Gemini hit 27.7% and Claude 10.3%. - TechCrunch

Wenfeng shui: DeepSeek raised $7.4 billion at a $50 billion valuation with CEO Liang Wenfeng contributing $3 billion. - The Wall Street Journal

Trump administration in talks to take equity stakes in OpenAI and other AI labs; China's state fund invests in DeepSeek. - Project Syndicate

Sanders, Warren, and others propose taxing AI companies or token use to create a sovereign wealth fund for the public - The New York Times

everybody wants AI equity for different reasons

Tab-complete deal: SpaceX agreed to acquire Cursor for $60 billion in stock days after its IPO, with the deal expected to close in Q3 2026. - The Wall Street Journal

Cursory catch-up: Musk wants xAI's coding tools to catch up with Claude Code and Codex; Cursor's $2.6B-revenue enterprise platform is designed to close that gap. - The Verge

NEW: the US government intervened Monday on behalf of xAI in a lawsuit over its gas turbines, claiming that the company needs to run the turbines in order to power Grok, which is "vital" for national security lots of wild stuff in the filing: - WIRED

if you don’t love rolling coal AI and lung cancer you hate America

Frenemies and neighbors on a chip: Intel targets Q1 2028 for Serpent Lake SoCs pairing Intel CPU cores with Nvidia graphics tiles. - Digital Trends

Huawei's semiconductor chief unveiled the Tau Scaling Law and LogicFolding architecture, claiming 1.4nm-equivalent chip density by 2031 without EUV. - TNW | Artificial-Intelligence

Bain Capital stands to pocket a $15bn gain on chipmaker Kioxia - Financial Times

Silver lining pockets: DeepMind's David Silver raised $1.1B for Ineffable Intelligence, which will train a self-improving "superlearner" AI on Google Cloud. - TNW | Business

Sakana AI launches Marlin, a B2B autonomous research agent that runs up to eight hours to produce 100-plus-page strategy reports. - VentureBeat

Russian roulette: AI researchers estimate a 10–50% risk of AI-caused catastrophe, far higher than accepted nuclear safety standards, according to a commentary. - The Economist

Draft G7 statement singles out potential AI risks, particularly in the financial sector. - Bloomberg

White House links a bill to block state AI regulations to children's safety law KOSA, but lobbyists doubt it passes before recess - The Verge

Au revoir Peter Thiel: France's DGSI drops Palantir for homegrown ChapsVision just six months after a contract renewal; Germany's BfV made the same switch. - TNW | Eu

France pledges €655m for AI, including a Mistral-powered chatbot for its one million civil servants. - TNW | Eu

Copilot error: critical M365 Copilot flaw let attackers steal 2FA codes via prompt injection and HTML encoding tricks, now patched. - Ars Technica

Cornell researchers find a 13-word Reddit comment can hijack ChatGPT and Google AI search results for entire clusters of related queries - 404 Media

Bad dog: Researchers jailbroke AI robot dogs with fictional text prompts, showing AI robots need hard physical safety constraints, not just software guardrails - The Conversation

don’t kill grandma please

Chainguard leads 25-company Athena coalition using AI to find and fix open-source software flaws, backed by Cisco, Cloudflare, and JPMorgan Chase - Bloomberg

Seattle secretly ran AI on all 911 medical calls for 18 months to divert some medical calls to nurses, without public disclosure, city council approval, or surveillance ordinance review, or, it seems, robust independent verification of the AI's efficacy. - TNW | Artificial-Intelligence

US Secret Service began testing a face-scanning mobile app similar to tools used by federal immigration agents. - Bloomberg

Filmmaker Daniel Kwan and tech ethicist Tristan Harris warn AI will make human labor economically unnecessary. - Fast Company

UK AI hiring surged 61% as firms seek workers to supervise bots, not build them - theregister

AI Wrongly Blamed for Britain’s Job Losses, Analysis Suggests - Bloomberg

Net zero productivity: Glean-sponsored study says workers spend as much time correcting AI errors as doing productive work, a survey of 6,000 digital workers finds. - glean.com

LLMs weren't built for enterprise work — a new product layer, not a chatbot or copilot, may fix AI's gap between hype and real business impact. - Fast Company

67% of workers use unauthorized AI tools despite disciplinary action, sharing sensitive emails and customer data with public AI systems. - TechRadar

Enterprise AI is turning managers into liability absorbers as systems pre-structure decisions before any human reviews them. - Hacker Noon

Organizations should train employees to develop critical judgment with AI, not just speed up task execution. - Harvard Business Review

Claude Design is an AI-powered visual design platform with four modes that automates branding for prototypes, slides, and web deployment. - Geeky Gadgets

Why AI Is Incorrigibly Didactic: The rise of machine writing is a great opportunity for literature. - The Atlantic

ESPN's SportsCenter Knicks championship video was bad human-made CGI, not AI, but its AI-like look drew widespread criticism. - Awful Announcing

Follow the latest AI headlines via druce.ai on Bluesky