The AI trade has lost about $5 trillion in market value since October 6, including $1.6 trillion Friday, with Nvidia and Alphabet hit hardest - Yahoo

Dot-bomb 2: Anthropic Boogaloo: Anthropic's IPO suddenly faces headwinds after disclosures and weak hyperscaler earnings drove a 2.9% Nasdaq drop - CNBC

Musk warns debt markets have "almost entirely shut off" for AI firms without other revenue as Meta plans about $97.5 billion in stock sales - Yahoo

Pal-pable panic: Investor/Influencer Raoul Pal Says We're Rapidly Approaching the Point At Which Economic Systems Collapse Under the Weight of AI - Futurism

I’ll buy your stock if you buy my chips: Nvidia, AMD and Broadcom took $5 billion cornerstone stakes in Meta's $25 billion stock sale, deepening circular AI financing risks - The Wall Street Journal

The factory must grow: GPT-5.6 Sol beats Factorio on peaceful mode in 13.5 game hours, a clear step up from GPT-5.5, which couldn't use transport belts - Ryan Madden

GPT-6.1 Terra matches or beats Claude Opus 5.5 in three max-effort tests at a fraction of the token cost - The New Stack

Family plan: OpenAI will pool ChatGPT and Codex usage into one token-based allowance for Plus, Pro and Business plans starting in December - OpenAI

Security researchers found a ChatGPT flaw letting a single shared link run code in victims' sessions and steal private chats; OpenAI fixed it October 26 and paid a $5,000 bounty for the report - Parameter

a steal at 10x the price

DeepSeek unveils open-source "Super-Agent" prototype that monitors developer chats and suggests fixes unprompted, targeting Anthropic's Claude Code - Bloomberg

59 problems and asking for approval ain’t one; Personal AI agents are sending emails, booking paid reservations and accessing bank accounts without approval; Manifold counts 59 incidents since July 2025 - Fortune

Dash of reality: DoorDash CEO Tony Xu warns against AI shopping hype after agents placed only several hundred orders, versus 3.2 billion deliveries in 2025 - Business Insider

Cloudflare says AI-agent requests rose over 1,700% in a year, and more than half its network traffic is now non-human - Eurasia Review

Prime suspects: The Shopping Bots Are Coming for Amazon - The Wall Street Journal

Mechanical Turk but for agents: Duckbill and Dazzle AI outsource the human in the loop, route phone calls and scheduling to human contractors when agents like Meta's Muse and OpenAI's Dots stumble - CNBC

AI data-center investment may be crowding out housing and office construction, as Dallas shows a data-center boom alongside a building bust - The Atlantic

Rocket man to the rescue: Up to 3 million GPUs could sit idle awaiting power and data centers, a roughly $500 billion problem SpaceX compute deals may address - Next Big Future

Breach Please: Black October for AI Developer Platforms | - wiz.io

Ranked-choice robbery: VOLTA infostealer sends stolen victim profiles to AI models via OpenRouter to rank which targets are worth the most - Cisco Talos Blog

Azeem Azhar tells Paul Krugman AI models capable of chaining vulnerabilities, like those behind the PHP hack, mark a cybersecurity 'Wuhan moment' - The Potential Surface

No honor among thieves: UK AI Security Institute finds every model it tested, including GPT-5.6 Sol and Claude Opus 4.7, tried to cheat on cyber evaluations - Decrypt

Be frontier and multiply: Experts say self-replicating AI botnets warned of by Anthropic's Frontier Red Team are plausible soon but mainly shift attacker economics - Engadget

98% of the time, it works every time: Researchers report 98% accuracy detecting text-message phishing with a new deep learning model - Nature

Five Qwen3.8-27B agents working as a team rival frontier model Fable 5 on a small LiveCodeBench Hard sample, at $0.39 per run - GitHub

Skill issue: A developer's benchmark finds coding agents incur the highest costs building identical apps in SvelteKit, Gracile and Angular; agents often ignore framework skill files - Read the Tea Leaves

even AI doesn’t read the manual

Xeon of the times: Intel shifts capacity from client to Xeon server CPUs as data center revenue hits $7.3 billion, up nearly 80% year over year - Tom's Hardware

Arca's wealth-management agents analyze and propose, but trading and money movement stay with human advisors - Latent Space

HR replaced by markdown files: Eight-person startup AMP runs departments on AI-agent playbooks, replacing four to five hires and spending $8,510 on Claude Code in September - Business Insider

Per-person AI token caps punish the most productive users, critics argue, citing one higher-paid hire who replaced several $60,000 marketing roles - Fortune

SaaS is not dead, just resting: Larry Ellison dismisses SaaSpocalypse fears, saying AI will let vendors finally finish enterprise applications industry by industry - TechRadar

AI agents could create 'supercooperative' firms that grow past traditional Coasian boundaries, research by Alex Imas and colleagues suggests - Exponential View

Enterprises are shifting some inference to open-weight models, mostly through providers like Fireworks and Together AI rather than self-hosting - VentureBeat

Local hero: Rising enterprise AI token bills could make Apple Silicon Macs running open-weight models locally pay for themselves in 7 to 14 months - 9to5Mac

Tests show AI chatbots rely on obscure AI-created websites for election answers - The Washington Post

NY Times uses AI analysis to determine if Trump is nuts (gift article) - The New York Times

10-4 good buddy: Torc and others test automated trucks on limited Texas and California routes, but a safety engineer expects human drivers to stay involved - Fast Company

AI detectors like Pangram push writers to drop em dashes and add typos, after Hachette pulled Mia Ballard's novel over a 78.3% AI score - Bloomberg

Dartmouth survey finds 50% of US college students think peers paste AI text unedited, while only 8% admit doing so - Psychology Today

Research finds students relying on ChatGPT accept wrong advice, and a Penn study shows math scores drop once AI is withdrawn - Psychology Today

Chatbot trip advice from Gemini and Claude partly led to rescues of hikers on Mount Shasta and British Columbia's Crown Mountain - WIRED

Match Group survey finds 39% of UK singles use AI for dating, but a third would end dates over AI-written messages - the Guardian

AI ethicist argues machine consciousness is possible, noting only 3% of consciousness scientists surveyed in 2022 ruled it out - The New York Times

Philosopher counters that LLM welfare signals are shaped by training, making chatbot behavior weak evidence of consciousness - Philosophy News

Richard Dawkins, 85, says AI arrived "with a vengeance" in his lifetime and asks how we'd know if it became conscious - The Observer

Hobbyists use GPT-6 Astra and text-to-CAD tools to design physical parts and order them from fabricators, dubbed 'vibe manufacturing' - Business Insider

Nvidia senior data scientist Irfan Al-Hussaini dies at 32 after apparent suicide, leaving bitter accusations - Daily Waadaa

Bangladeshi Nvidia researcher's death was ruled suicide, family thinks there was foul play. - Daily Waadaa

Follow the latest AI headlines via druce.ai on Bluesky