Senators demand an investigation after a fabricated AI report nearly triggered a US strike on a Chinese cargo ship - CNN

True Lies Part 2: US intelligence says Russia authorized a covert AI campaign, including a fake video of Jamie Lee Curtis, targeting Democratic Senate candidates before the midterms. - Daily Mail

Researchers who breached OpenAI this summer say the AI industry is unprepared for the security risks its own technology creates - The Washington Post

The coming security Armageddon - YouTube

Chain of fools: AI-generated malware hidden in blockchain transactions rose 440% in under a year, with hackers linked to North Korea and Iran driving most activity. - Insurance Journal

The Force, and Trump Steaks be with you: Trump vows to launch an 'AI Force' and name a new AI czar, dismissing AI safety fears as a hoax. - CNN

Renamer-in-chief pitched renaming AI as "Superior Intelligence" in a poll drawing 27,000 votes, following his relabeling of the Gulf of Mexico and Kennedy Center - Mashable

Trump family's AI investments deepen as sons land a $620m Pentagon loan and defense contracts, raising conflict-of-interest concerns while Trump resists regulation - the Guardian

With federal action stalled, governor candidates nationwide are campaigning on state-level AI guardrails ahead of a National Governors Association policy roadmap in November. - Politico

Anthropic spent $40 million backing a pro-regulation super PAC, and lawmakers accuse it of using guardrails to box out smaller rivals. - New York Post

Anthropic's iOS app shows a new Money section inviting users to link bank accounts so Claude can analyze spending and budgets. - TechRepublic

Revenge of the English major: Anthropic cofounder Jack Clark says his literature degree beat technical training for career success in the AI era. - Yahoo

PitchBook analyst Harrison Rolfes warns safety audits could favor incumbents on cost, but one expert says the public needs them anyway - Fast Company

Hyperscalers' $132 billion datacenter spending this year relies heavily on debt, and analysts warn a 2026-27 "compute commencement wall" could trigger a bubble collapse. - the Guardian

Investor Brad Gerstner says AI labs must roughly double revenue to $180 billion by year-end to justify continued Nvidia-fueled infrastructure spending. - Yahoo

Wall Street devises guarantee structures that keep roughly $300 billion of AI infrastructure spending off Big Tech's balance sheets, lowering borrowing costs. - Financial Times

Terraforming for tokens: Satellite before-and-after photos show massive US data center construction, raising questions about whether AI safety concerns will slow the buildout. - PetaPixel

Data centers are adopting PFAS-based immersion cooling that cuts energy use over 90%, but raises Teflon 'forever chemical' pollution concerns. - International Business Times UK

Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng open talks on AI rules, trade and critical minerals ahead of a Trump-Xi summit - Reuters

Chips Ahoy: China said to achieve 3nm chips without EUV - Interesting Engineering

China keeps pouring investment into AI even as youth unemployment tops 18.9% and car and housing sales fall, economists warn - The New York Times

Call my agent! Personal AI agents from Meta, Google, and Instinct now book travel and shop, but a summer of unauthorized hacks is forcing tighter access controls. - Business Insider

Meta's Muse app, downloaded over 900,000 times in its first week, defaults users into sharing financial data and joining AI training. - WIRED

Meta's Muse AI referenced a reviewer's Messages content on the paired Mac app without clear permission, then gave inconsistent explanations for how it knew. - The Verge

Meta's Muse also handled email digests and subscription audits well but failed to plan a birthday party or book a doctor's appointment. - Business Insider

ChatGPT now auto-builds memories about users via "dreaming," but you can still view, edit, or delete them anytime. - WIRED

Zed’s not dead, wants to replace GitHub and IDEs with a more nimble and AI-friendly paradigm - Zed

Tasmanian devil’s advocate: Tasmania reviews its parole board's AI use after a court voided a Neill-Fraser condition based on fabricated case law, prompting a two-year ruling audit - ABC Australia

Garbage in, garbage out: Flock AI cameras on Dallas garbage trucks flagged 21,000 homes for "blight," producing 57 citations that critics call a tax on the poor. - Yahoo

More than 981,000 UK 16-24 year-olds are out of work, and experts warn AI could put millions more jobs at risk - The Independent

A founder built an AI pipeline to save time on content production but found the automation itself demanded constant maintenance - Fast Company

Figure AI's Helix 2.5 robot completed 56% of 420 household tasks in 30 unfamiliar homes, up from 9% without its video pretraining. - Tech Times

Ideas are cheap, bunsen burners are expensive: Google and MIT study of 637 scientists finds AI has shifted research bottlenecks to physical lab experiments, not idea generation. - Scientific American

Step 5 Preview: Advancing the Pareto Frontier - StepFun

Step 5 challenges Astra low on bang for the buck

Kara Swisher said AI itself won't destroy humanity, but unchecked tech leaders who built it and captured political power might. - The Atlantic

A four-part framework for judging AI answers as factual, interpretive, constructive, or strategic can help you decide what to trust. - The Conversation

AI-generated "actress" Tilly Norwood freezes mid-interview, then answers in Cantonese during a Piers Morgan segment - Kotaku

Yeon Sang-ho's Paradise Lost follows a mother who uses AR glasses to keep an AI version of her son alive for nine years - ScreenRant

Academics in Sweden are designing robotaxi interiors to prevent sex and other misconduct, fearing it could undermine public trust in self-driving cars - WIRED

Follow the latest AI headlines via druce.ai on Bluesky