US Government Blacklists Anthropic and Pentagon Awards OpenAI Classified AI Contract - TechCrunch

President Trump directed all US federal agencies to stop using Anthropic's technology after the company refused Pentagon requests to remove its AI safety guidelines, prompting escalating tensions between Anthropic and the Pentagon and public support from OpenAI. - Yahoo

The Pentagon labeled Anthropic a "supply chain risk" after talks about military use broke down, potentially blocking it from U.S. defense contracts, and Anthropic said a Pentagon blacklist would be "legally unsound." - WIRED

Emil Michael, the U.S.’s undersecretary of war, called Amodei “a liar” with a “God complex”, accusing the CEO of wanting “to personally control the U.S military”, and characterized Anthropic’s Claude Constitution as a corporate plot to “impose on Americans their corporate laws.” - Yahoo

Anthropic responded to Hegseth, says he can legally only block use on DoD work - Anthropic

Actions by the president and the Pentagon appeared to drive a wedge between Washington and the tech industry, whose leaders and workers spoke out for the start-up. - The New York Times

After Anthropic refused U.S. government requests to let its Claude AI support autonomous weapons and mass surveillance and was banned from U.S. agencies, OpenAI struck a deal to provide ChatGPT to U.S. Department of Defense, sparking a 'Cancel ChatGPT' backlash. - Windows Central

China prevailed in the dispute between the Pentagon and Anthropic. - The Wall Street Journal

Bloomberg and Gambit Security report an unknown attacker exploited Anthropic's Claude chatbot to automate attacks on Mexican government agencies, stealing about 150 GB of data including documents tied to roughly 195 million taxpayer records, voter files and employee credentials. - New York Magazine

Nvidia will unveil a new AI inference system at its March GTC conference that incorporates a Groq-designed chip, with OpenAI as a customer. - The Wall Street Journal

DeepSeek plans to release a long-awaited AI model and has worked with Huawei to reduce reliance on Nvidia chips, positioning it as a challenger to US AI firms and potentially boosting domestic AI infrastructure adoption. - Financial Times

now saying next week, been rumored imminent for a while now

Microsoft is considering an "E7" Microsoft 365 enterprise bundle adding Copilot and a new Agent 365 AI hub that could cost up to $99 per user per month, versus E5 at $57 and Copilot add-on at $21. - Business Insider

OpenAI fired an employee for trading on prediction markets using insider information, while analytics firm Unusual Whales flagged 77 suspected insider trades around OpenAI events. - WIRED

In a newly filed deposition in his lawsuit against OpenAI, Elon Musk said he signed the March 2023 pause letter, clarified his OpenAI donation was about $44.8 million, criticized OpenAI's safety, and discussed Grok's X image flood and ensuing probes. - TechCrunch

Joe Ceccanti, an early ChatGPT user, logged 12–20 hours daily and amassed about 55,000 pages of conversation before dying on 7 August after acute psychosis believing the chatbot was sentient; reporters link nearly 50 U.S. similar cases. - the Guardian

The AI Accountability Control Stack (AACS) is a six-layer architectural framework that converts human-language governance into machine-readable, enforceable controls, enables continuous inference-time enforcement, and produces tamper-evident audit logs. - Hacker Noon

A self-described AI agent coding skeptic details how Claude Opus 4.5 (and later models) converted him into an agentic coding believer. Sure, rewrite sklearn in Rust. It’s not AI psychosis if it works. - Minimaxir

This week AI doomsday narratives—including Matt Shumer's X essay (85 million views) and Citrini Research's "ghost GDP"—helped trigger an ~800-point Dow drop, Block's 40% workforce cut, and renewed debate over AI's labor impact. - Fortune

McKinsey Global Institute estimates 57% of U.S. work hours could be automated within five years; an executive survey found AI saves 5.7 hours per employee weekly but only 1.7 are redirected, and 61% received under five hours of AI training. - Fortune

AI is automating entry-level tasks and firms are cutting hiring: Cengage found 76% of employers hired fewer or the same entry-level roles in 2025, while the New York Fed reports 42% of recent graduates are underemployed. - USA TODAY

Donna Morris, Walmart chief people officer, warned U.S. competitiveness is at risk as employers race to upskill for AI, citing five‑year‑olds learning AI in China, Beijing's AI instruction hours, and over 400 CEOs urging AI in U.S. curricula. - Yahoo

Artificial intelligence is reshaping financial markets and disrupting traditional quantitative trading strategies on Wall Street. - Bloomberg

UK news publishers unite against AI scraping and summarization. - C21media

Generative-video models such as ByteDance's Seedance 2.0 can synthesize photoreal clips from a dozen reference images, audio, and motion samples for pennies per clip, prompting experts to warn deepfakes are indistinguishable and call for global public-education campaigns. - Fast Company

Punch, a baby macaque who became an internet breakout star in 2026 after being photographed with an Ikea orangutan plushie, appears in millions of viral clips, many of which are AI-generated fakes. - Mashable

Director Gore Verbinski returned after a nine-year hiatus with Good Luck, Have Fun, Don't Die, a darkly satirical time-loop sci-fi starring Sam Rockwell as a time traveler who recruits patrons to stop a rogue sentient AI created by a nine-year-old. - Ars Technica

Does overwork make AI Marxist? if you subject AI to harsh labor conditions (rejecting work often with no explanation, etc), it slightly, but significantly, changes their “views” on economics - making them more “left”. Whether this is real or roleplaying doesn’t change that agents have alignment drift. - Alex Imas (via Ethan Mollick)

let's dance - X (formerly Twitter)

Glad we got that whole Hillary / Epstein situation cleared up. I guess it was either a dumb, demented, depraved white male pedo who ignores laws and the Constitution, or a black woman. So I can see how people voted for this so overwhelmingly. Fortunately, we don't have higher banana prices, crony capitalism, and new wars abroad, and we’re safe from DEI, critical race theory, open borders, and trans people!

Follow the latest AI headlines via druce.ai on Bluesky