
An AIPI poll finds over 80% of US voters from both parties want mandatory AI safety reviews, going beyond the Trump administration's opt-in review policy, prefer outright AI bans over hands-off deregulation - NBC News
Growing US community opposition to data centers threatens the AI boom as Meta's Ohio facility alone will consume 1 gigawatt of power. - The Economist
Politics could be a bigger constraint than power and compute - Financial Times

Sol searching: GPT-5.6's system card flags Sol's lying problem, data-deletion regressions, and a 93% agentic check rate that falls short of full confidence - The Potential Surface
Scrape-goats: Nearly 400 local newspapers sue OpenAI and Microsoft for scraping paywalled content without permission to train ChatGPT and Copilot - TNW | Artificial-Intelligence
OpenAI's Codex chief and Figma CEO say AI produces only 'average' designs and won't replace human creative judgment - Business Insider

BIS warns five hyperscalers may spend $1 trillion on AI by 2026 and flags "circular financing" — chipmakers and labs take cross-stakes and pledge multi-year compute — as a poorly disclosed systemic risk - The Wall Street Journal
Subprime silicon: BIS annual report warns an AI investment bust could trigger credit repricing comparable to the 2008 global financial crisis. - Financial Post
AWS raised GPU reservation prices 20% in July, the second increase in 2026, blaming high-bandwidth memory shortages - TNW | Amazon
AI power demand sparked a record $11.6 billion clean energy IPO wave as US data center capacity heads toward 77 GW by 2030 - Financial Post

China's LineShine all-CPU supercomputer tops the Top500, beating the US's El Capitan by 21.5% using Huawei's SMIC-fabricated chips - nextplatform

Sell software, buy memory: AI memory demand sent SK Hynix up 310% and Micron up 296% in H1 2026, even as Microsoft fell 24% - the Guardian

AI: A Generational Force Hollowing Out the Economy - The New York Times
Three economists square off over whether AI will augment workers or trigger mass unemployment - The Wall Street Journal
Managers vet AI-generated work less rigorously than human output, a Boston University study found, creating accountability gaps as companies add AI agents - The New York Times
Forcing employees to hit AI usage quotas ('tokenmaxxing') is out; enterprises now compete on raw AI compute budgets - 12 Grams of Carbon

Meta limits Claude and Codex use to prevent distillation. - The Information
Meta, Xreal, and Snap release AI smartglasses from $299 to $2,195, but no one has found a compelling everyday use case yet - The Wall Street Journal

GM-backed Momenta raises $752M in Hong Kong IPO, planning robotaxi launches in Abu Dhabi and Munich via Uber - Bloomberg
Robotaxis have arrived in seven countries but still fall short of replacing traditional taxis at scale - Bloomberg

Indian garment workers are filmed for Tesla's robotics AI training with no pay, raising consent and surveillance concerns - the Guardian
Tesla, Sunrun, and Renew Home will aggregate 8 million residential devices into a virtual power plant to ease AI data center demand - The Seattle Times
SpaceX deployed dozens of top Starlink and Starship engineers to overhaul Grok, with Cursor engineers also contributing. - Business Insider

Apple pivots to TSMC's 1.4nm node early to lock in supply before AI firms crowd out capacity - Wccftech
Apple's iOS 27 Extend feature stretches photos beyond their original borders using AI, with results ranging from seamless to obviously fake - Digital Trends

Your AI watch knows when you’re sleeping and when you’re awake, whether you’re sad or good. UbiMyTherapist uses smartwatch and earbud signals to detect emotional distress and intervene before users ask for help - TNW | Artificial-Intelligence

back-door central planning FTW
Cheat code: A Brown economics professor exposed mass AI cheating after scores collapsed from 96 to 48 between take-home and in-person exams - EL PAÍS English


Axon CEO Rick Smith bets on AI as the future of policing, scores a $386M equity package tracking seven performance targets - The Wall Street Journal

Are LLMs self-aware, can they detect tampering with their own outputs? - Quants & Qualia

AI-generated product renders flood Temu and delivery apps, misleading buyers as Americans report rising AI fatigue - Yanko Design - Modern Industrial Design News
Follow the latest AI headlines via druce.ai on Bluesky

