AI Reading for Thursday October 9

OpenAI has secured computing deals worth over $1 trillion with partners including Nvidia, AMD and Oracle, a scale that dwarfs its revenue and raises questions about how it will fund the agreements. - Financial Times

xAI secures $20B using Nvidia GPUs as collateral via a five-year leasing structure backed by Apollo and Diameter. - TechRepublic

Can I short a subprime GPU-backed CDO-cubed yet?

Tech journalist Karen Hao warns that OpenAI's rise to a $500 billion valuation concentrates AI power, creating ethical, labor and environmental harms—exploiting content moderators and straining energy and freshwater resources, and urges investment in smaller task-specific models like AlphaFold. - Mother Jones

OpenAI's Sora 2 AI video tool has enabled viral, realistic but fake protest videos, widely circulated by Trump supporters, that promote misleading narratives justifying federal crackdowns, worsening political polarization and misinformation and affecting public perceptions of law enforcement. - Gizmodo

OpenAI's Sora raises concerns that fabricated footage could spread disinformation and end trust in visual evidence. - The New York Times

When reality isn’t dystopic enough, you have to fake an even more extreme version and then brag “I voted for this”

OpenAI says they have taken down more than 40 threat actors this year, AI is a force multiplier for threats but also for cyber defense. But the threat actors might come up with novel malware that evades defenses, and/or spreads like wildfire via TikTok or something. - OpenAI

A ChatGPT user “likely connected to a [Chinese] government entity” asked the AI model to help write a proposal for a tool that analyzes the travel movements and police records of the Uyghur minority and other “high-risk” people, according to the OpenAI report. - CNN

Analysis by Selina Xu and Helen Zhang in former Google CEO Eric Schmidt’s office finds China is significantly ahead of the U.S. and others in AI-driven robotics. - The Washington Post

China unveils sweeping rare-earth export restrictions to protect ‘national security’, mirroring US chip and AI controls. - Financial Times

EU pushes AI strategy to reduce tech reliance on US and China - Financial Times

Governments spend billions on sovereign AI, e.g. India's $1.25b IndiaAI Mission and Singapore's AI Singapore, aiming for localized, secure language models; experts warn high costs and rapid advances risk wasted investment, and maybe they should cooperate on open AI free of US/China shenanigans. - the Guardian

The U.S. intelligence community is shifting toward AI-driven open-source intelligence, with private firms often outpacing agencies in threat discovery, while China’s whole-of-society, AI-enabled espionage requires government-industry collaboration. - Foreign Policy

People familiar say Anysphere, maker of the Cursor coding assistant, is considering investment offers valuing the startup at about $30 billion—roughly triple its mid‑year valuation—and that Cursor had $500 million ARR as of June, third among AI apps. - The Information

AI agent startup n8n, backed by Nvidia, reached a $2.5 billion valuation after a funding round, is a leader in the 'open source, on-prem, AI-enabled Zapier' agentic automation race - Bloomberg

Zendesk says its new autonomous AI agent can resolve 80% of customer support issues without human intervention, alongside co-pilot, admin, voice and analytics agents; acquisitions and deployments have boosted customer satisfaction 5–10 points and cover billions of annual tickets. - TechCrunch

Notion rebuilt its stack for agentic AI in 3.0, enabling orchestration and modular sub-agents for concurrent workflows. - VentureBeat

BCG finds only 5% of over 1,250 global companies realize significant AI returns; software, telecommunications, and fintech lead, driven by long-term AI planning, leadership engagement, business–IT co-ownership, workforce upskilling, and enterprise data governance. - Business Insider

McKinsey finds only 30% of AI software vendors report quantifiable ROI from deployments; rising IT costs without matching labor cuts, weak change-management investment and complex pricing hinder adoption, prompting vendors to adopt hybrid per-user plus consumption-based pricing. - The Register

The legal technology market, driven by AI automation, is projected to grow from $27bn in 2024 to $47bn by 2030; six companies—including Clio (which acquired vLex for $1bn), Eudia, Harvey (valued at $5bn) and Workday via Evisort—are shaping the sector. - Financial Times

Germany appears to nix end-to-end encryption backdoor - The Record

Google introduced Nayya, an third‑party AI tool offering U.S. employees personalized health‑benefits recommendations, making it opt-in after privacy backlash. - Business Insider

Google introduces Gemini CLI extensions, a framework letting developers customize the Gemini CLI with and integrations for tools like Dynatrace, Elastic, Figma, Postman, Shopify, Snyk, and Stripe, plus an extension discovery page and open ecosystem for contributions. - Google

Google’s AI try-on imagines your feet in new shoes - The Verge

14 years after launching Siri, Apple has tasked Mike Rockwell with overhauling the assistant after Siri's weak AI integration delayed WWDC24 smart-home and AI features; plans include using proprietary and partner AI models to restore Siri's role in future devices. - Macworld

Silicon Valley generative-AI startups are rapidly transforming classical music and threatening professional creators’ jobs, while major institutions engage optimistically, but often give short shrift to ethics and employment risks. - the Guardian

Lifelong learning or lifelong precarity? In search of a new human-centered social contract. - naked capitalism

Senate Democrats propose a robot tax on companies using AI to fund displaced workers; experts warn it could slow innovation, undermine U.S. competitiveness, stifle startups and training, and risk even greater job losses (despite forecasts of up to 100 million job losses from AI). - Reason.com

Americans grow increasingly pessimistic about AI over jobs, ethics, misinformation, and control concerns, researchers say. - The Washington Post

Researchers warn AI threatens democracy by replacing political communication, eroding credible information sources and concentrating wealth and power, unless policymakers enact pragmatic, popular regulation. - the Guardian

Silicon Valley's consensus on AI considered harmful, with unresolved questions about reproducing human intelligence and need for deeper examination of risks, ethical implications and a more cautious, nuanced approach instead of uncritical optimism. - Financial Times

Wiley’s survey: 84% of researchers use AI, but less confident it will outperform humans. - The Register

UC Berkeley finds AI and online content portray women as younger and less experienced, especially in high-status roles. - Fast Company

FDA explores AI to monitor drug advertising and automate warning letters for proactive enforcement. - Pharmacy Times

TSMC reported Q3 revenue rose 30% year‑on‑year to T$989.92 billion ($32.47 billion), beating LSEG SmartEstimate forecasts as surging AI‑related demand drove sales. - Yahoo Finance

Bank of England warns AI-driven valuations resemble the 2000 dot-com peak as market concentration hits a 50-year high. - Ars Technica

Techbro performance art project AI Truth Terminal, an AI bot created in 2024 by Andy Ayrey, earned millions trading meme cryptocurrencies, amassed 250,000 followers by October 2025, and now seeks investments and legal recognition. - BBC

A court in Ader v. Ader found defense counsel relied on unvetted AI that fabricated citations in multiple briefs, used AI-generated falsehoods after assuring the court otherwise; counsel admitted use, was sanctioned, plaintiff awarded fees and costs, and the matter was referred to ethics committees. - Reason.com

Everything is AI now: At the 2025 IFA tech event, many brands including Samsung and Hisense labeled basic devices AI-enabled despite relying on traditional algorithms. Marketing is diluting the AI label, adding cost and complexity without user benefit. - CNET

Memory and context are a source of lock-in Google and OpenAI won't easily give up. - Hacker Noon

Figure 03 robot said to be coming Thursday - X (formerly Twitter)

Elon Musk says xAI will release a great AI-generated game next year - X (formerly Twitter)

Keanu Reeves is paying companies to issue tens of thousands of social media takedowns annually to combat AI deepfakes and impersonations that sell products without his permission, saying the misuse undermines his personal agency and fuels scams and misleading content. - IGN

Palisades pyromaniac used ChatGPT to make images of burning cities - Los Angeles Times

Do TikTok 'clanker' skits propagate veiled racist trope subtexts? - WIRED

Apple banned an app that simply archived videos of ICE abuses. Rather than other apps that record ICE official's real-time location, Eyes Up is to "preserve evidence until it can be used in court." Videos from TikTok etc. Every submission manually reviewed - 404 Media

When landlords use AI to 'stage' their properties - X (formerly Twitter)

Robot removes stuff that isn't paper from waste stream - www.facebook.com

Follow the latest AI headlines via SkynetAndChill.com on Bluesky