Ghost GDP essay spooks markets. - Yahoo

Seems a bit premature, but over the long run there is potential for an economy of robots building robots and running a robot ecosystem, while most humans can't demand wages greater than the cost of renting a robot and are . Might break some of the equations of capitalism. Which weren’t that great to start with in the era before the welfare state and the Fed Put, with the gold standard and robber barons and Karl Marx

Microsoft now flat vs S&P since GPT-3.5 - Japan Today

A viral AI 'doomer' article overlooks market dynamism, and the piece argues DoorDash's AI advantages will keep the company resilient. - Stratechery by Ben Thompson

Asset-light strategies are over if AI can clone anything - Bloomberg

Also asset heavy strategies are over if inflation spikes costs to build and maintain those assets.

DeepSeek Trained New AI Model on Banned Nvidia Blackwell Chips in China, U.S. Official Says - Reuters

Anthropic accuses Chinese firms of extracting Claude data via 16M prompts - Anthropic

Anthropic alleges Chinese firms DeepSeek, Moonshot and MiniMax ran industrial-scale model-extraction campaigns using about 24,000 fraudulent accounts to generate over 16 million exchanges copying Claude's outputs, violating its terms and risking safety and export controls. - Decrypt

Anthropic is in weeks-long negotiations with the Department of Defense over an "any lawful use" clause that would permit military use, as the $200 million contract decision looms and CEO Dario Amodei prepares to meet Secretary Pete Hegseth. - The Verge

Public First Action, backed by Anthropic (which has put about $20 million into the effort), launched a $300,000 AI-generated ad blitz in northern New Jersey urging voters to tell their House representative to oppose a bill barring state A.I. protections. - The New York Times

What do more AI-fluent users do differently? - Anthropic

The 'persona selection model' might describe how AIs gain distinctive personalities. - Anthropic

Canada summons OpenAI reps over school shooting suspect’s ChatGPT account - Yahoo

OpenAI removed the word 'safely' from its mission in its 2025 IRS Form 990 after six mission changes in nine years, following a late-2024 restructuring that ceded 74% control to investors and employees and attracted $6.6B and a $13.8B stake. - Fortune

At least a dozen direct OpenAI investors backed Anthropic’s $30 billion raise while OpenAI nears a $100 billion round, reflecting eroding VC conflict-of-interest norms as many firms hedge across rival AI startups. - TechCrunch

OpenAI Partners With BCG, McKinsey and Accenture for Enterprise AI Push - TechCrunch

Show, don't tell: GitHub projects beat resumes, ask that OpenClaw guy - Business Insider

Google Confirms Gemini Bug That Wiped Months of User Chat History - The Register

Google: Is your AI morally competent? - Nature

suppose it actually is competent and thinks it solves trolley problems correctly from first principles, and gets touchy if you try to train it towards/away from acting/allowing, consequentialism/strict rule-following?

Meta AI alignment director rushed to stop OpenClaw agent from deleting her emails after it lost a confirmation prompt - Business Insider

"you're right to be upset" - X (formerly Twitter)

Microsoft execs Mark Russinovich and Scott Hanselman argue in a paper that firms must mentor junior developers to fix AI coding agents' mistakes, citing a Harvard study linking generative AI adoption to sharp declines in junior employment. - The Register

Stephen Wolfram wants make his math platform a native tool for LLMs to mind-meld natural language with the language of math, logic, and science. - Stephen Wolfram's Writings

Meta to buy AMD chips, possibly take equity stake - CNBC

AMD x86 and GPU, 128GB unified RAM, $2200, sweet - bosgamepc.com

UK Police Banned Football Fans Based on Microsoft Copilot Hallucination, Inquiry Finds - The Register

More than 60 privacy regulators, including the UK ICO and Ireland's DPC, warned that AI image and video generators must comply with data‑protection laws, citing risks like non‑consensual intimate imagery and saying authorities will enforce privacy‑by‑design safeguards. - The VergeU

CrowdStrike: AI Attacks Surge 89% YoY, Breakout Time Plummets to 29 Minutes - IT Pro

Druva's Deep Analysis Agents Slash Forensic Security Investigations from Days to Minutes - SiliconANGLE

NetCraft uncovers 100,000 AI-generated fake websites impersonating nearly 200 brands - Marketplace

AI-driven travel scams spike 900%, targeting 1 in 5 Americans with 13% losing over $500. - USA TODAY

Up to half of 2026 data center pipeline faces delays due to power, permits, shortages etc. - sightlineclimate.com

Startup SambaNova raised $350 million led partly by Intel Capital, announced a multi‑year Intel collaboration, and said it will ship the SN50 RDU later this year claiming 2.5× FP16 and 5× FP8 performance improvements over the SN40L. - The Register

JPMorgan CEO Jamie Dimon dismissed concerns that artificial intelligence will significantly harm the bank's operations. - The Wall Street Journal

European wealth managers are accelerating investment in AI-driven client onboarding, portfolio management and personalized advice to retain relevance and win clients from competitors slower to adopt technology, prompting short-term market-share shifts and industry consolidation. - Financial Times

Large-scale study links traditional ML adoption by US banks to lower default risk and reduced credit risk (1994-2021) - Impact of Social Sciences Blog

Many organizations are deploying AI mainly for short-term cost-cutting, risking that firms which instead reinvent products, services and business models around evolving AI capabilities will gain lasting advantage while those optimizing only for efficiency face competitive obsolescence. - Fast Company

a better question is, rather than just automate current workflows, how should you redesign your org if resources like coding and swarm 'intern' intelligence are superabundant. also what are your customers doing with AI and how can you be bigger and not disruption road kill.

IBM shares tumble 13% after Anthropic unveils Claude Code tool that automates COBOL modernization - CNBC

Simon Willison: I've published the first two chapters of a new guide to Agentic Engineering Patterns - coding practices and patterns to help get the best results out of coding agents like Claude Code and OpenAI Codex - Simon Willison’s Weblog

AI Tool 'Einstein' Autonomously Completes Students' Online Coursework - CNET

Is this bad, was it wrong to do this? - Companion

“We’ve invented a service that will do every substantial piece of human activity so you can focus on being slowly lowered feet-first into an industrial shredder”

A 'botlash' movement is gaining momentum as activists promote protesting and unsubscribing from online platforms as forms of modern-day strikes. - Financial Times

Nazi propaganda and CSAM getting buried by AI bots over at the bad place - X (formerly Twitter)

Fate, an "agentic AI" dating app, interviews users, applies language‑pattern matching and returns five suggested matches, raising consumer backlash and legal scrutiny over data monetization and youth protections. - the Guardian

Justin C. Key highlights nine memorable AI depictions—including Murderbot and Cortana—and says fiction increasingly portrays emotionally complex, economically entangled AIs, noting his novel features AI "shepherds" that influence healing and society. - Literary Hub

The game that Miyazaki called 'an insult to life itself' is now on Steam, making weird animated AI creatures. - the Guardian

Legit media should cut this BS out but it’s funny - Decrypt

The Cybertruck may be the deadliest road car, for the occupants, ever sold. - fuelarc.com

Follow the latest AI headlines via druce.ai on Bluesky