Defense Department sources explain how generative-AI chatbots could analyze and rank targets for human vetting, adding a conversational layer to Maven tool. - MIT Technology Review

The Pentagon increased reliance on AI-enabled weapons systems, prompting alarm among some personnel. - Bloomberg

The U.S. military used Anthropic's Claude embedded in Palantir's Maven Smart System to support real-time targeting and prioritization, reportedly enabling strikes on about 1,000 targets in the first 24 hours, while Claude acted as decision-support, not an autonomous weapon. - Fast Company

Palantir demos and Pentagon records show AI chatbots, including Anthropic's Claude, being tested to analyze intelligence and generate or suggest war‑planning options, raising questions about reliability, oversight and governance before deployment. - WIRED

U.S. and Israeli Operation Epic Fury struck roughly 4,000 targets in the first 100 hours (~1,000/day), reportedly killing over 1,300 civilians including about 175 children at Minab, while the Pentagon employs AI tools such as Palantir's Maven and LLMs. - The Intercept

A Pentagon dealmaker has become a nemesis of AI company Anthropic. - The Wall Street Journal

Anthropic's Pentagon Dispute Highlights AI Procurement Risks for Federal Contractors - The Wall Street Journal

Anthropic doesn’t trust the Pentagon, and neither should you. - The Verge

OpenAI accepted a Pentagon contract after Anthropic refused surveillance uses, prompting user and employee backlash including a nearly 300% rise in ChatGPT uninstalls, and CEO Sam Altman amended a memo forbidding intentional domestic surveillance amid critics citing ambiguous contract wording. - Techdirt

Anthropic says widespread AI deployment could enable an "AI panopticon" with extensive surveillance and centralized control. - Bloomberg

The 1993 X-Files episode 'Ghost in the Machine' depicts a corporate-created sentient AI sparking a CEO-versus-DoD safety conflict that parallels modern disputes, including Anthropic versus the Pentagon, over military AI use, procurement friction, and legal and reputational risk. - Fast Company

Ukraine's defense ministry will make millions of drone videos and battlefield datasets available to Ukrainian and allied firms, managed by its Defense Ministry innovation center and already used to train the domestic "Delta" system, while retaining human lethal-targeting authority. - The New York Times

ByteDance accesses Nvidia's top AI chips via Southeast Asian data centers, despite US export restrictions - The Wall Street Journal

Groq and roll all night at GTC: Nvidia is expected to unveil an inference-focused chip using Groq technology with OpenAI as a likely customer, plans Feynman with copackaged optics, reported roughly $6 billion in robotics revenue last quarter, and halted H200 shipments to China. - Business Insider

Nvidia to unveil Rubin GPUs, Vera CPUs, Feynman architecture, new inference chips, and OpenClaw software at GTC 2026 - The Register

A cottage industry is developing around OpenClaw, installing it and supporting it and providing solutions on top of it - MIT Technology Review

OpenClaw adoption turns China into a massive real-world testing lab for autonomous AI agents - Bloomberg

Prompt like a hard-ass boss who won't tolerate failure and your bot swarms will find ways to hack your system and breach policies. - The Register

Running WebAssembly inside a transformer. - Percepta

if you can get transformers to natively run tools within a single inference call, like fetching data or solving an equation, and train LLMs to do that, it takes them to a whole new level , a huge step closer to AGI

Venture capitalist Hemant Taneja and other Silicon Valley leaders acknowledge an AI bubble as Big Tech plans about $650 billion in AI spending this year, with OpenAI valued larger than Toyota, Coca-Cola and Disney combined. - The Atlantic

Sam Altman said intelligence should become a metered utility, while OpenAI cited rising compute and data-center energy costs—canceling a planned Stargate expansion in Texas. - Gizmodo

utilities get regulated tho

Big Tech spending drives surge in AI researchers leaving academia for industry - Schneier on Security

Avocado toast not well done: Meta delays rollout of new model, better than previous models but doesn't match latest biggest Gemini, Claude etc. - The New York Times

Lovable CEO said artificial intelligence will break software monopolies. - Bloomberg

Alex Karp says AI is good for electricians and plumbers, bad for pink collars. - HuffPost

i’m not sure this is controversial motivated thinking

Musk unveils Macrohard project pairing xAI Grok with Tesla AI agent on proprietary chips - CNBC

Ex-xAI: xAI has lost cofounders as Zihang Dai left and Guodong Zhang said he will leave, leaving only two of the 11 original engineering cofounders remaining amid a Musk-led reorganization, layoffs, and IPO preparations. - Business Insider

xAI hired senior Cursor leaders Andrew Milich and Jason Ginsberg, and Elon Musk said he expects xAI to catch up with rivals in coding by the middle of this year. - The Information

Elon Musk personally vets xAI candidates following co-founder departures - Business Insider

Google and Samsung announced Gemini task automation, beginning a beta rollout on the Galaxy S26 Ultra that automates app tasks like rideshare and food delivery by navigating UIs, asking clarifying questions and requesting final user confirmation. - The Verge

Google launched Ask Maps, a Gemini-powered conversational layer in Google Maps available in the US and India that uses Gemini to analyze reviews, photos and past Maps interactions to answer complex, personalized real-world questions and provide directions. - The Verge

NYT feature on programmers using Claude Code - The New York Times

AI-assisted coding exposes a split among developers between craft-focused programmers and those who prioritize directing AI to get results, shifts work toward AI engineering and reviewing outputs, and raises consolidation and open-web concerns. - blog.lmorchard.com

Claude can now respond with interactive visuals - Claude

Claude for Excel and PowerPoint Updates - Claude

now a single context can understand your whole task involving Excel and PowerPoint

MIT hosted the 2025 NSF-funded Workshop on the Future of AI+MPS and published a Machine Learning: Science and Technology white paper recommending coordinated investment in computing and data infrastructure, cross-disciplinary methods, and workforce training. - MIT News | Massachusetts Institute of Technology

Math, Inc.'s reasoning agent Gauss translated Maryna Viazovska's Fields Medal–level proofs of the 8- and 24-dimensional sphere-packing problem into fully verified Lean code, completing months of verification in days and boosting formal-proof confidence. - Live Science

AI helps unblock physicists on complex calculations, moving from coder to collaborator - The Economist

Adobe CEO Shantanu Narayen, after 18 years, will step down once a successor is appointed and remain board chair, the company said, as shares fell over 7% and Adobe reported Q1 revenue of $6.40 billion with Q2 guidance $6.43–6.48 billion. - Yahoo

Ford launched Ford Pro AI at Work Truck Week, making it available to all U.S. Pro telematics subscribers; built on Google Cloud, it analyzes fleet data (seatbelt use, fuel, health, idle times, speeding) and Ford Pro exceeds 840,000 subscribers. - Yahoo

STMicroelectronics will retrain staff and deploy over 100 humanoid robots in older fabs over the next couple of years to handle repetitive, physically demanding tasks and avoid closures amid a restructuring that proposed 5,000 departures in October 2024. - Reuters

Amazon added an adults-only 'Sassy' Alexa+ personality that can use explicit language and roast, requires extra security checks (e.g., iOS Face ID), is disabled with Amazon Kids, and, Amazon says, won't generate explicit sexual content, hate speech or self-harm. - TechCrunch

Amazon to lay off another 14,000 employees in Q2 amid aggressive AI transition - technext24.com

Alexa, am I getting fired today?

Angela Lipps, 50, was misidentified by Fargo police using facial-recognition in a bank‑fraud probe, arrested July 14, extradited Oct. 30 and spent over five months jailed before release Dec. 24 after bank records showed she was 1,200 miles away. - Grand Forks Herald

Autocomplete my (lack of) personality: Cornell-led researchers found that AI autocomplete suggestions can sway users' beliefs and change their expression and tone as such tools become widely integrated into emails, surveys and mobile keyboards. - Scientific American

The AI adoption gap: How differing tech habits are straining modern personal relationships - The Washington Post

It’s not you, it’s your AI

Hollywood is being reshaped by layoffs, consolidation, streaming losses and AI, prompting questions about whether the film industry has hit a temporary rough patch or is facing terminal decline. - Bloomberg

Big Tech moguls are moral opportunists with tunnel vision, you say? - The New York Times

this is my surprised face

Getting a marketing email from a rogue AI…AI ain't putting up with your crap

Follow the latest AI headlines via druce.ai on Bluesky