Broligarch pro-AI Super PACs aligned with Silicon Valley are mobilizing ahead of the US midterms, escalating spending and using AI-driven targeted outreach to influence elections and protect commercial AI business models while shaping upcoming AI regulation. - WIRED

White House launched a public-relations campaign rebranding AI as the 'Trump Revolution' promoting AI-driven economic growth to counter public fears of job loss, reassure markets and accelerate AI adoption while shaping regulatory and labor-policy debates. GDP is coming back! but robots are doing all the work. - The Washington Post

Salesforce CEO Marc Benioff, speaking at the World Economic Forum in Davos, urged regulation of AI after warning large language models had 'become suicide coaches' and referenced several documented suicides while calling to reshape Section 230. Is the suicide rate actually going up, and is it because people have to use Salesforce. - CNBC

Geoffrey Hinton said he was sad at what his life’s work had become, warned advanced AI could outsmart humans, resist shutdown and pose existential risk, predicted job losses and unrest, and urged urgent research and international regulation comparable to chemical and nuclear controls. “AIs can afford to be careless, but not men” said godfather of AI. - Business Insider

At Davos 2026, Nvidia's Jensen Huang, Microsoft’s Satya Nadella, Anthropic’s Dario Amodei and DeepMind’s Demis Hassabis warned of uneven AI deployment, geopolitical risks, and safety concerns, with Amodei saying AI could eliminate up to half of entry‑level white‑collar jobs. - euronews

At Davos, AI dominated panels and sponsor houses from firms such as Palantir, Workday and C3.ai. CEOs reported amazing AI outcomes (Aramco $3–5 billion savings), while President Trump’s arrival dominated side conversations, injecting geopolitical tension ahead of 2026. - MIT Technology Review

AI and Trump have taken over Davos, which was dominated by AI and big-tech companies showcasing branded storefronts and hospitality centers, shifting the forum toward commercial, market-focused agendas from prior years’ energy transition, trade. - New York Times

Palantir CEO Alex Karp told the World Economic Forum that AI will shift most work to vocational and manufacturing roles, destroy humanities jobs, and let rich countries halt immigration. - Futurism

put down the books, learn to drive a forklift, until they are self-driving

Yuval Noah Harari told the World Economic Forum in Davos that AI will create two crises for every country: an identity crisis as AI outperforms human thinking and an "immigration" crisis as AI displaces jobs and imports skills. - Business Insider

Anthropic CEO Dario Amodei said at Davos that AI could produce 5–10% GDP growth and 10% unemployment, warned about ~10 million people capturing outsized gains up to 50% GDP, and urged government action to manage displacement. - The Wall Street Journal

the 10/10/10 economy, 10% growth, 10% unemployment, 10 people hogging the wealth

IMF says AI could boost growth up to 0.8% but is "hitting the labor market like a tsunami" (lifeboats sold separately). Challenger, Gray & Christmas linked ~55,000 U.S. 2025 layoffs to AI, including Amazon's 15,000 and Salesforce's 4,000 cuts. - CNBC

CEOs say AI is making work more efficient, while employees report it has not improved their work experience. - The Wall Street Journal

At the World Economic Forum in Davos, Microsoft CEO Satya Nadella said AI sovereignty depends on who controls models and embedding a firm's tacit knowledge into model weights, calling datacenter location "the least important thing" and citing encryption as mitigation. - The Register

Translation, you can't trust Microsoft anywhere in the world because Trump can order them to cut off your email b/c you said something he doesn't like.

Microsoft CEO Satya Nadella warned AI must "do something useful" or lose social permission to use scarce energy, urged workers and firms to adopt AI as a cognitive amplifier, and cited a report finding 95% of organizations get no return. - PC Gamer

Ukraine will share nearly four years of wartime combat data, including millions of hours of drone footage, with allies to train military AI models, while defence minister Mykhailo Fedorov launches a performance-based, data-driven overhaul of the defence ministry. - Reuters

who owns the training data, owns the future

TSMC said it had invested too little to meet overwhelming AI demand, creating capacity constraints, supply and pricing pressure for AI chips and prompting calls for increased investment, policy support and alternative foundries to diversify supply. - Stratechery by Ben Thompson

OpenAI said its Stargate data centers will pay their own energy costs, minimize potable water use for cooling through design and AI, and work with communities by securing dedicated power or funding local grid upgrades. - The Verge

New analysis finds AI-driven growth in data-center computing will drastically increase U.S. power-plant emissions and electricity demand over the next decade, potentially raising prices unless data centers shift to renewables and improve energy efficiency. - WIRED

Energy use of a day with Claude Code is like running the dishwasher an extra time. - Simon P. Couch

we promise to only gently roast the grid

At Davos, Amazon CEO Andy Jassy said the AI market may be a bubble, cited circular dealmaking and heavy compute use, and said Amazon is investing in AI infrastructure including Project Rainier's $11 billion Georgia datacenter and Trainium chips. - The Register

Amazon CEO Andy Jassy said on Tuesday that AI-powered shopping could significantly boost Amazon's retail business, that Amazon already has a shopping chatbot, and that the company is exploring hosting third-party chatbots on its site. - The Information

Hildebrand said the AI bubble issue is critical to the inflation view. - Bloomberg

Meta's Superintelligence Labs delivered its first high-profile AI models internally this month, CTO Andrew Bosworth said at Davos, calling them "very good" while warning substantial post-training work remains before public release. - Reuters

ready to start undermining mental health and reality at scale

Humans&, a three-month-old AI startup with about 20 engineers from OpenAI, Anthropic and Meta, raised a $480 million seed round at a $4.48 billion valuation led by SV Angel and Georges Harik, with participation from Nvidia, GV and Jeff Bezos. - SiliconANGLE

OpenAI's Ending the Capability Gap report found power users use seven times more advanced "thinking capabilities" than other users, and OpenAI will expand its OpenAI for Countries initiative in 2026 to scale adoption across education, health and other sectors. - OpenAI

OpenAI has begun deploying an age-prediction model that infers whether ChatGPT accounts belong to users under 18 from behavioral and account signals, will roll out in the EU in coming weeks, and will enable stricter safety settings for flagged accounts, and let OpenAI deploy ‘spicy’ capabilities. - The Register

OpenAI says it will unveil its first ChatGPT-powered, behind-the-ear earbuds in the second half of this year, with leaks targeting September 2026 and plans for up to five devices by Q4 2028. - CNET

can talk to the voice in your head all day

Documents from Elon Musk’s lawsuit show Microsoft executives negotiated expanded funding and pushed monetization beast mode as OpenAI moved to a capped for‑profit in March 2019, launched paid ChatGPT in Jan 2023, and grew revenues tenfold from 2023–2025. - Gizmodo

AI companies are integrating ads into conversational search/browser interfaces, with OpenAI launching ChatGPT Search in late 2024 and ChatGPT Atlas in October 2025, and Google earning over $1.6 trillion in ad revenue since 2001. - Fast Company

Kuaishou's Kling AI goes ‘chaching’, reached over 12 million monthly active users, according to a source. - The Wall Street Journal

China AI firm Zhipu limited access to its coding agent after demand for the service surged. - Bloomberg

Anthropic publishes take-home exam. "If you optimize below 1487 cycles, beating Claude Opus 4.5's best performance at launch, email us at [email protected] with your code (and ideally a resume) so we can be appropriately impressed and perhaps discuss interviewing." - GitHub

Using Claude Code with a local model - Ollama

go llama shepherd cottage-core

Clawdbot, an open-source LLM-powered personal agent by Peter Steinberger, runs locally, integrates messaging apps and devices, accesses the filesystem and shell to run commands and automations, and supports multiple models plus TTS and image generation. - MacStories

Wilson Lin at Cursor ran hundreds of autonomous coding agents for nearly a week to build a web browser, producing over 1 million lines of code across about 1,000 files and trillions of tokens, yielding a mostly working but glitchy app. - Simon Willison’s Weblog

Hacker News posters share success and complaints from using "agentic coding" to build an iOS SwiftUI pet-feeder app that produced many bugs, duplicated and fabricated code, and ineffective guardrails, warning it creates technical debt and requires architectural sign-off and human review. - Hacker News

A former Amazon executive's 'vibe coding' post describing a 'wildly productive weekend' sparked debate about viral AI tools. - GeekWire

gets you to ‘mostly working’, speeds development but you have to give it specifics and tests and iterate and check everything.

Claude Code is a huge deal for people who can't code, the way podcasts are a huge deal for people who can't read - YouTube

The Agentic AI Handbook catalogs 113 production-ready agentic patterns across eight categories, attracted about 2,500 GitHub stars over Christmas 2025, and reports pattern results like 2-3x success for Plan-Then-Execute and 10x+ speedups for Swarm Migration. - nibzard

113 ways to leave your developer… move fast and vibe things

A test of six free AI chatbots — ChatGPT, Google Gemini, Microsoft Copilot, Claude, Meta AI and Grok — found every model hallucinated at least once, including invented cases, erroneous image IDs and fabricated book titles, posing legal and reputational risks. - ZDNET

Check Point Research discovered VoidLink, a vibe-coded cloud-targeting Linux malware framework uploaded to VirusTotal Dec. 4 that contains about 37 plugins and 88,000 lines of AI-generated code produced in a wildly productive six days week, likely authored by a single developer using Trae Solo. - The Register

MedHELM provides a clinician-validated taxonomy (5 categories, 22 subcategories, 121 tasks) and a 37-evaluation benchmark to assess LLMs; in head-to-head tests across nine models, advanced reasoning models (DeepSeek R1, o3-mini) won about 66% of comparisons. - Nature

Tell me about your motherboard: AI therapy apps are widely filling gaps in Italy's scarce public mental-health services, with a survey showing 81% view mental health as a weakness, 57% cite cost as the main barrier, and about five million cannot afford support. - the Guardian

Admissions vibe-checked by the algo: Students will get their college decisions faster this year because of artificial intelligence being used by admissions departments to read essays and review transcripts. - Bloomberg

An AI-infused warehouse sorts real Louis Vuitton bags from fakes. - The Wall Street Journal

AI for VLOOKUP: Datarails raised $70 million in a Series C led by One Peak, bringing total funding above $175 million, launched AI-native, Excel-first FinanceOS conversational agents, and reported revenue growth over 70% year-over-year with headcount topping 400. - SiliconANGLE

Company deployed an AI layer that automated root-cause analysis and playbooks, raising adoption from ~20% to ~100% and reducing churn 10%, with a target 25% reduction by H1 2026. - Hacker Noon

AI wrote your resume, but can you do a pushup, or understand the Constitution: Border Patrol recruiters at a St. Petersburg recruitment event used an AI resume generator to rapidly create and sometimes submit applicants' resumes, producing tentative job offers but resulting in unvetted candidates who later failed fitness tests and forced reassignments. - Raw Story

Google DeepMind's CEO said robotics is nearing a physical-AI breakthrough. - Bloomberg

Elon Musk said Tesla's Cybercab robotaxi, slated to begin production in April, and the Optimus humanoid, targeted for production by end of 2026, will see agonizingly slow initial ramps despite goals of 2 million Cybercabs and 1 million Optimus annually. - Business Insider

Fireside chat with xAI engineer turns into actual firing, as he dishes on extreme engineering culture. - YouTube

Elon Musk and Sam Altman trade barbs about AI safety - X (formerly Twitter)

Musks attack seems a bit rich when MechaHitler Pedo Grok lacks guardrails and undresses women and kids, while Teslas keep burning people alive]

Some pretty good robot dance moves (no Charleston). If it's real (and I have doubts), better balance and coordination than I've seen. - YouTube

the robots gonna steal the viral videos and deepfake the ‘Dum Dummies’

In China, husbands have offered up to RMB20,000 (about $2,800) to persuade spouses to stop playing the viral ‘otome’ dating sim Love and Deepspace, whose AI voice-chat, AR companion and ChatGPT integration are more attentive than real spouses and have been linked to relationship breakdowns. - International Business Times

Bring a Trailer listed a 1999 Cadillac DeVille with multiple AI-generated or altered images flagged by users, then withdrew the auction within hours and apologized, with company executives David Duke and Howard Swig announcing an internal review. - Jalopnik

During AAA's Fox Latin America debut, WWE aired an AI-generated promo showing Dominik Mysterio with an AEW-logo belt, prompting AAA to delete a tweet and WWE to remove the video package. - Dexerto

3 years of AI video progress - Reddit

Every time you enter an AI prompt, a complex sequence of stuff happens - Reddit

Follow the latest AI headlines via druce.ai on Bluesky