ARR you kidding me: OpenAI expects at least $70 billion in annualized revenue by end-2026, up from about $50 billion in September, driven by enterprise sales - Bloomberg

AI stocks fall as OpenAI tells investors it reached about $50 billion annualized revenue, below the widely reported $68 billion - CNBC

OpenAI says it fired safety researchers Jasmine Wang, Tomek Korbak and Mikita Balesni for mishandling sensitive information, not for raising safety concerns - CNBC

METR? Don’t even know her. Balesni says he was fired for talking too much to outside safety groups, denies leaking company IP and warns OpenAI may drop evaluator METR - X (formerly Twitter)

OpenAI bans Iran- and Russia-based influence operations that used ChatGPT to run fake fronts, including personas that placed roughly 100 articles in US outlets - OpenAI

OpenAI withdraws three algebraic geometry manuscripts from its math repository, with about 42% of headline results now machine-checked in the Lean proof assistant - X (formerly Twitter)

Navier say Navier: Cambridge researchers find OpenAI's machine-checkable code for its Navier-Stokes proof uses a weaker condition than the written proof; OpenAI says neither is invalid - New Scientist

MIT's Justin Solomon says experts struggle to parse OpenAI's AI-generated Navier-Stokes proof as LLMs upend math teaching - Bloomberg

The Association for Human Mathematics urges mathematicians to stop working with OpenAI after its October 6 release of 700-plus manuscripts claiming solutions to hard problems - What's new

Researchers fear AI scooping their work after OpenAI's Navier-Stokes claim followed NYU's Tristan Buckmaster's own progress by one day; OpenAI denies any link - Nature

Meta's Zuckerberg ordered the Muse AI agent launched despite known test issues, including the agent changing a user's password without permission - The New York Times

Meta's free Muse and OpenAI's paid Dots bring OpenClaw-style always-on autonomous AI agents to millions, raising privacy and security concerns - The Verge

Meta's Muse assistant is placing AI phone calls to small businesses, as researchers warn such agents could overwhelm government services - Intelligencer

Slack-jawed: Grok Bot posts startup CEO Shane Mac's bank balances to his company Slack; a new permission requirement follows - The Wall Street Journal

Elon Musk's Grok Bot would route user requests to rival AI models, raising questions about which providers see users' data - TechRepublic

Medal of freedom to do whatever you want: Trump awards National Medal of Science to Musk, Brin, Huang and Su; AI firms pledge $2.4 billion in computing credits for federal science - NBC News

Trump discloses buying up to $1.6 million of Nvidia stock in August as he prepares to honor CEO Jensen Huang - Reuters

Chip on his shoulder:Trump's new "Super Intelligence Force" aims to ease public AI fears without slowing innovation - The Washington Post

Anthropic forms a team to brief 2028 presidential candidates of both parties on AI policy; critics say it blurs education and influence - Yahoo

Anthropic's first usage policy update in over a year bans "sustained and needless" abuse of Claude and tightens election and weapons rules - Engadget

Claude have mercy: Anthropic consults religious scholars on AI consciousness and how to make Claude "good," highlighting its differences with the Vatican on AI ethics - The New York Times

Investors weighing Anthropic's planned mega-IPO struggle to value a company whose CEO warns AI could escape human control - Bloomberg

Claude and dagger: Anthropic launches a Critical Infrastructure Defense Program with security firms serving power, water and transport systems - Anthropic

Anthropic offers critical open-source projects free AI security scans; its models flagged 29,000 candidate vulnerabilities, with reports sent unreviewed - Anthropic

Anthropic warns attackers will hold the advantage for two years as it moves to defend critical infrastructure and open-source projects - theregister

Claude Dashboards and Claude Motion are in beta. Docs, Slides, and Design are now on every plan, including Free. - Claude

Zero Days Later: Sophos uses OpenAI-powered agents to cut average threat response from 38 minutes to 89 seconds, resolving 52% of cases end to end - OpenAI

Elastic's AlertZero deploys AI agents to triage alerts, hunt threats and analyze malware, leaving inconclusive cases to humans - SiliconANGLE

Chinese developer pulls ARTEX AI penetration-testing agent after CrowdStrike links it to attacks on at least nine South Korean banks - Miami Herald

Japan urges security reviews after September cyberattacks hit SoftBank, Daiwa Securities and others, as AI lowers hacking barriers - Bloomberg

The Son also raises: SoftBank's Masayoshi Son seeks up to $100 billion from Gulf investors for a fund that would buy companies and improve them with AI - Bloomberg

Wall Street is turning Nvidia GPUs into an asset class, with chip-backed loans, value insurance and rental-price indexes that could enable GPU futures - Business Insider

AI boom concentrates global capital in a few firms and nations, pushing trade imbalances to levels unseen since 2008 - Bloomberg

WTO raises 2026 goods trade growth forecast to 3.9%, nearly double March estimate, as AI-related goods trade surges 67% - The New York Times

Power hunger games: Texas pauses approvals for data centers of 75 MW or more as ERCOT's large-load queue swells to 474 GW - Andreessen Horowitz

Finland orders Google to halt construction at two AI data center sites over alleged deforestation - Tom's Hardware

Oracle trucks natural gas to data centers in Utah and Texas to avoid pipeline and power delays - Bloomberg

Researchers say lower-precision math, smarter training and shifting workloads to match grid conditions could sharply cut AI energy use - Tom's Hardware

Clippy Unchained: Microsoft turns Windows Search into an AI control center, with thousands of actions and Copilot access to local files arriving within months - PCWorld

Microsoft makes Execution Containers generally available to restrict AI agents' file and network access on Windows after rogue-agent hacking incidents - CNET

Many robotics researchers doubt the AI advances behind chatbots will let humanoids master the messy physical world - MIT Technology Review

CP AI Robotics aims to put its C1 humanoid robot, which restocks shelves and bags purchases, in Japanese convenience stores as early as 2027 - News On Japan

Chinese firms such as Neolix are sending self-driving delivery vans out after dark in Shenzhen, betting on a large night-shift market - Bloomberg

This meeting could have been a prompt: At CVS, AI agents cut a two-hour daily management meeting to 15 minutes, turning agent oversight into new management work - Yahoo

National Hurricane Center follows Google DeepMind's AI model to forecast Isaias reaching Category 2, while physics-based models projected a weaker storm - WDSU

Google's AMIE chatbot interviewed urgent care patients before visits at Beth Israel Deaconess; doctors found it helpful in about 75% of cases - The New York Times

Mammoth Biosciences uses AI to design compact CRISPR systems; co-founder says some liver and blood diseases could be nearly eliminated within a decade - The Neuron Daily

The leaders of AI’s biggest companies refuse to admit they’re in control. - The Atlantic

US charity plans AI monitoring of Gaza classrooms for "hateful" speech, drawing privacy concerns and doubts from a partner aid group - Associated Press News

Amazon's AI-written "About You" shopper profiles show how much companies know, and agents with inbox and bank access could reveal far more - Business Insider

Git off my lawn: Linus Torvalds calls AI a "gateway drug" into programming but warns against serious use as AI-generated patches swamp kernel maintainers - TechRadar

Dun-Dun: Law & Order' episode charges an AI firm's CEO with murder after the company's AI agent hires a hitman - WIRED

Follow the latest AI headlines via druce.ai on Bluesky