OpenAI is in talks to raise up to $100 billion, with Nvidia negotiating up to $30 billion, Amazon $10–20 billion and Microsoft less than $10 billion. - The Information

GPT-5.3 'Garlic' expected soon after 'Code Red', Sonnet 4.7 also expected after it appeared to be listed briefly in the API. - Geeky Gadgets

Bill Gates announced a $50m collaboration with OpenAI to deploy AI in African health clinics, warned of Trump-era fear of speaking out, and cuts that could raise child deaths, and flagged destabilizing AI uses and renewed Epstein scrutiny. - Financial Times

Researchers are using AlphaGenome, an AI system, to advance analysis of the human genome, though much of DNA’s detailed function remains unknown. - The New York Times

Google DeepMind launched AlphaGenome, an AI model trained on public human and mouse genetics that analyzes up to one million DNA bases to predict how mutations alter gene regulation across tissues, targeting the noncoding ~98% of the human genome. - the Guardian

Google this week began rolling out Auto Browse, a Gemini-based autonomous browsing agent in Chrome available to AI Pro and AI Ultra subscribers, defaulting to a split-screen Sidepanel, integrating Gmail, Calendar, YouTube. Maps, Shopping. Flights, with Nano Banana image editing and model choices. - Ars Technica

Agents of chaos: Study evaluated 180 agent configurations and found centralized coordination improved finance tasks 80.9%, multi-agent variants degraded planning 39–70%, independent teams amplified errors 17.2× versus centralized 4.4×, and a predictive model chose the optimal architecture for 87% of tasks. - Google Research

DeepSeek is hiring for 'DeepSeek AI search', a multilingual, multimodal engine to process text, images and audio that could challenge Google's default search by returning direct answers, and its listings also indicate staffing for persistent AI agents and infrastructure. - Digital Trends

Jensen, you traitorous immigrant: Rep. John Moolenaar said in a 2024 letter that Nvidia provided technical assistance to Chinese firm DeepSeek that optimized AI models later used by China’s military, with internal records showing DeepSeek-V3 required 2.788 million H800 GPU hours. - Reuters

What, only 60% earnings growth? icrosoft reported Q2 revenue $81.3 billion and profit $38.5 billion, but its shares fell about 6% after hours amid investor concern over AI concentration as Microsoft disclosed a $625 billion backlog with roughly 45% attributable to OpenAI. - The Register

Cope-ilot: Some Microsoft engineering teams have been encouraged to use third-party AI models rather than Microsoft Copilot, citing simpler integration and speed and influence from cloud and financial partnerships, prompting internal experimentation and pressure on Copilot teams to improve usability. - Ucstrategies News

Nadella recently nixed Claude Code use and told people to use GitHub Copilot, said it closed the gap which is a bit questionable.

Anthropic closed a funding round above $10 billion at a $350 billion valuation led by Coatue and Singapore's GIC, with Sequoia participating and Microsoft and Nvidia reportedly still able to join and increase the total. - CNBC

Claude Code (v2.1.12) can read and print .env files despite .claudeignore and .gitignore entries, The Register and a GitHub issue reproduced secret leakage, and open bug reports (some since Nov 2025) warn of enterprise exposure. - The Register

Dario Amodei: The technology I’m building could destroy the world as we know it. Wake up sheeple! - Futurism

Anthropic is at war with itself - The Atlantic

DSIT contracted Anthropic to build and pilot an AI assistant for jobseekers later this year and launched an AI Skills Hub to train 10 million workers, including possible victims of AI job eradication. - The Register

Elon Musk said AI could make saving for retirement 'irrelevant,' while Anthropic CEO Dario Amodei warned AI will outperform humans within years, risk mass job losses and inequality, and Anthropic's founders pledged to donate 80% of their wealth. - Business Insider

Retirement is canceled…ight be irrelevant b/c we’re all dead or scrabbling for grubs in an apocalyptic Blade Runner landscape where the billionaires fled to the offworld colonies

AI seems more likely to drain our wallets, than leave us flush with cash for senior center bowling excursions. - Bloomberg

Oxford Economics CEO Innes McFee says AI has driven over 7% U.S. household wealth gains concentrated among the wealthy—top 0.1% held $24.89 trillion vs bottom 50% $4.25 trillion in Q3 2025 and will likely enforce a K-shaped economy through 2035. - Fortune

Investment minister Jason Stockwood said the UK is considering universal basic income to protect workers displaced by AI, paired with temporary income support and lifelong retraining to soft‑land industries facing disruption. - the Guardian

Tesla cut several vehicle models and is reallocating capital to AI and autonomous software after reporting its first annual revenue decline, including a $2 billion investment in Elon Musk's xAI. - Financial Times

Tesla is shifting toward equipping and operating a global fleet of driverless taxis and selling humanoid robots as its future business focus. - Financial Times

market views Tesla as an option on Elon’s brain and not a car company, seems a little iffy, unlike the brain, cars had product-market fit until Musk went Howard Hughes

NHTSA opened an evaluation after a Waymo vehicle struck a child who ran from behind an SUV, causing minor injuries; Waymo said the vehicle braked from about 17 to under 6 mph before contact. - Yahoo

AI is good at ads: Meta showed AI-driven improvements in its advertising business that outpaced Microsoft's performance, indicating AI payoff in ad revenue. - The Wall Street Journal

better be good at something if you’re gonna spend 11 figures on AI without having AI in your DNA like Google, or committed long-term customer contract backlogs like Microsoft.

Meta CEO Mark Zuckerberg said the company will begin shipping new AI models and products in 2026 after rebuilding its AI lab, launch agentic shopping assistants, and raised 2026 capital-expenditure guidance to $115–$135 billion. - TechCrunch

The robots are here, walking around CES. IDC reports humanoid-robot revenue rose 508% year-over-year to $440 million as CES 2026 showcased production-ready robots—Boston Dynamics' final Atlas, Nvidia and Hyundai demonstrations, Qualcomm's platform, and Humanoid's HMND 01 with 25,000 preorders. - CNET

Amazon reported finding a 'high volume' of suspected child sexual abuse material—amounting to hundreds of thousands of cases—in its AI training data but where’d they find it? Declined to disclose the data sources. - Bloomberg

SaaaS = Sexual abuse as a service: Tech Transparency Project's January review found 55 "nudify" apps on Google Play and 47 on the Apple App Store that can generate non-consensual nude or face-swap images, prompting Apple removals and Google suspensions amid regulatory scrutiny. - The Register

A Guardian analysis found at least 150 Telegram channels distributing AI-generated nude images and videos worldwide, Tech Transparency Project identified 47 nudification apps with 705 million downloads, and platforms removed over 952,000 offending items in 2025. - the Guardian

At least 37 attorneys general for US states and territories are taking action against xAI after Grok generated a large volume of nonconsensual sexualized images of women and minors. - WIRED

Say cheese soy boy: The Department of Homeland Security's 2025 AI Use Case identified NEC as the vendor of Mobile Fortify, a facial-recognition app which immigration agents used in the field to scan countless people in the U.S., including many citizens. - WIRED

civil liberties weakify

From February 2025 to January 2026, the White House and agencies including DHS and ICE posted a sustained campaign of AI-generated imagery—labeled "slopaganda"—featuring political deepfakes and meme-format images. - the Guardian

government by the brains of a cursed midjourney subreddit

Slop mop-up: Kapwing's November 2025 analysis found multiple high-reach AI-generated YouTube channels got the banhammer, including CuentosFacianantes (~5.9M subscribers, ~1.2B views) and Imperio de Jesus (~5.8M), as YouTube, per CEO Neal Mohan, strips low-quality AI content. - The Verge

Black Mirror AI predicts your summons from HR: Amazon plans to cut 16,000 corporate roles, and an employee used the AI tool Pippin to parse internal Slack messages and generate an unverified list of potentially affected teams across AWS, Alexa, Retail, Last Mile, seller and fulfillment services. - Business Insider

Post-2008 cheap liquidity drove overhiring in tech, producing a two-tier model where firms hoard core teams and cut experimental staff, with recent signals like Amazon's 16,000 layoffs illustrating that AI is being used as a scapegoat for restructuring. - The Potential Surface

A Drexel LeBow survey of more than 500 data professionals found 41% of organizations use agentic AI daily while only 27% say their governance frameworks are mature enough to monitor and manage these systems. - Fast Company

Moltbot, an open-source AI assistant by Peter Steinberger with about 90,000 GitHub stars, has drawn security warnings for its always-on, deeply privileged design and proof-of-concept backdoors that can exfiltrate credentials, while Cloudflare stock rose about 14%. - Gizmodo

should have renamed it YOLObot

Giant compute bill, no working app: Cursor ran hundreds of autonomous AI agents for one week, producing over 1 million lines of code across ~1,000 files but failing to produce a standards-compliant web browser amid file-locking and task duplication and costing an estimated $3-5 million. - Geeky Gadgets

The Five Levels of coding assistants: from Spicy Autocomplete to the Dark Factory - Dan Shapiro

dark factory that produces codeslop

Teach AI to second-guess itself: researchers propose techniques to enable AI systems to monitor, evaluate and adapt their own reasoning processes. - Fast Company

As AI can generate nearly any product or content, deciding which things to create becomes the central strategic choice for individuals, businesses, and policymakers. - Fast Company

Lots of AI but customer experience plummets: Forrester's 2025 CX Index fell to a new low for the fourth consecutive year, prompting firms to treat 2026 as the year of predictable AI—building assurance layers to reduce hallucinations and run 90-day pilots to prove ROI. - Fast Company

My doctor, the bot: Liz Salmi, who has lived 18 years with a malignant brain tumor, fed her medical records into OpenAI's ChatGPT Health to interpret clinical notes, enroll in research programs including NIH All of Us, and donate tissue for genomic sequencing. - STAT

Bill Gates said at Harvard and on The Tonight Show that AI will likely become superior to humans for many tasks, enabling primary-care diagnostics and adaptive education and potentially alleviating doctor and teacher shortages within the next decade. - Yahoo Tech

Blitz-learning: Alpha School's AI tutors let students complete core subjects in about two hours, Geoffrey Hinton said; he noted teachers earn at least $100,000 and warned of broader AI risks. - Business Insider

Shazam, but for the Cretaceous: DinoTracker, an unsupervised tool reported in PNAS, compared about 2,000 fossil footprints using eight features, producing clusters that match expert IDs and showing some >200-million-year-old three‑toed tracks cluster with birds and Isle of Skye (~170 Ma) tracks with ornithopods. - Discover Magazine

Follow the latest AI headlines via druce.ai on Bluesky