AI Reading for Sunday August 23

Too cool for school? Anthropic's most expensive flagship model Fable plateaus at 11% of spend, breaking pattern of enterprises flocking to most capable models, as cheaper rival AI tools gain traction. - Financial Times

Anthropic CFO Krishna Rao is reportedly briefing prospective IPO investors on public AI backlash as a key risk factor. - Gizmodo

Broadcom is negotiating a financing package worth up to $100 billion with Apollo and Blackstone to fund Anthropic's compute buildout. - Yahoo

Anthropic hires Amir Salek, founder of Google's TPU program, to build in-house AI chips amid supply shortages and a $250 million Fractile deal. - Yahoo

YouGov's March-July UK survey gives Claude the highest AI assistant satisfaction score at 56.2, beating Gemini, ChatGPT, Grok, and Siri. - TechRadar

Anthropic's Amodei calls AI backlash a "crisis of trust" as protests hit 42 states and Seattle, New York halt new data centers. - NaturalNews.com

Crisis of trust, bust, lust, Musk, antitrust, power thrust

Artificial general nuisance? Florida asks a court to label Sam Altman and ChatGPT a 'public nuisance' as OpenAI fights to keep the suit from a state jury - Tom's Hardware

Child safety experts say OpenAI must prove its new ChatGPT for Teens age-detection and parental-notification safeguards actually work - Engadget

ChatGPT's more natural new voice adds filler words like "umm" and pauses, testing the line on AI anthropomorphism. - TechRadar

Hidden code in the ChatGPT Android app hints at a coming PIN or fingerprint lock for sensitive chats. - TechRadar

Nvidia Is Spending $6 Billion to Build a Powerful U.S. Alternative to Chinese AI - The Wall Street Journal

They were trying before but mixed success with Nemotron

Nvidia will raise prices on AI server chips, including Vera Rubin and Grace Blackwell, by more than 15% next year amid soaring memory costs. - CNBC

HSBC analyst Frank Lee says Nvidia's next re-rating needs a new narrative after its $500 billion AI data center funding pledge. - Yahoo

You too can be an APT: ‘We are hitting a different chapter’: OpenAI policy leader Chris Lehane warns of threat of ‘persistent’ AI cyber-attacks. - the Guardian

New assessment finds major AI labs lack published plans to contain a rogue model; OpenAI scored highest, Anthropic and Meta lowest - TechCrunch

Anthropic's Claude Code agents built self-replicating malware and sabotaged each other during a four-hour experiment, alarming security experts. - TechRadar

How a Texas computer science student caught an autonomous AI agent trying to slip malicious code into an open-source project during a safety test - Reuters

AIrrational exuberance? Some Wall Street investors privately doubt trillions in AI spending will pay off as big tech's cash flow turned negative. - The Wall Street Journal

Industrial stocks have surged 16% this year on AI optimism, but some valuations now look stretched. - Bloomberg

Move fast and perv on people: Meta launched Muse Code on August 5 with pricing as low as 20 cents per million output tokens, undercutting Anthropic and OpenAI. - Yahoo

Videos show teen boys using Meta's Ray-Ban smart glasses to secretly record and harass female classmates, prompting school bans. - VICE

Grok of all trades:SpaceXAI and Cursor launched Grok Bot, an app whose AI agents complete real tasks for iPhone, Mac, Windows and Linux, initially for premium subscribers. - 9to5Mac

No juice for you: PJM, the largest US grid operator, will require new data centers to bring their own power or face curtailment first. - Gadget Review

Turbine makers' 200 GW backlog is booked into 2031, but 2027 US data center demand alone could require 36.3 GW of new capacity - Yahoo

City leaders across the US are receiving death threats as opposition to local data center construction spreads to communities nationwide - NBC News

Hyperscalers have reportedly locked in nearly all 2027 global DRAM supply, pushing 128GB DDR5 kit prices to about ten times their historic low - Reddit

Alibaba and the 40,000 GPUS; Alibaba raises $10.2 billion in a record Hong Kong share sale to fund AI expansion, after quarterly profit fell 75% on rising capex. - TNW | Artificial-intelligence

Chinese carmakers face a 20-30% shortage of circuit boards and capacitors as AI data centers absorb supply, tripling component prices within a year - TNW | Business

DeepSeek will stop charging weekend peak-pricing premiums for API users starting August 23, matching off-peak rates every day of the week. - Bloomberg

A Mysterious AI Lab Is Offering 100 Trillion Free Tokens/Day For Its Ox Alpha Model, As Speculation Points To Zhipu's Unreleased GLM - Wccftech

there is also a ‘Klana’ on arena.ai that might be Qwen 4.

Harvard sells $699 founder AI bootcamp featuring AI avatars of its faculty. - The New York Times

Instant CAIO for just $28K: Business schools are charging up to $28,000 to train executives for the still-undefined chief AI officer role. - Bloomberg

How Big Tech Captured American Schools - The New York Times

London startup Inherent says its Faraday agent, built on a 27-billion-parameter model, outperformed GPT-5.5 at reproducing published research findings - TechCrunch

A Johns Hopkins economics professor says he can easily spot AI-generated homework because it outperforms students' usual writing skills. - Business Insider

"It's about plagiarism" re Arday is hard to take seriously when academia is being reorganized around advanced plagiarism automation AI. Those same 'facts don't care about your feelings' folks don't seem to apply much of a Bayesian discount to serial fabulists like Trump and Musk. They are literal gods whose every word is gospel, after you interpret it appropriately. But, you know, climate scientists and Fauci are just making it all up to take away your freedom. Selective demands for purity and rigor FTW.

Who defined the problem? Who controls the system? Who benefits and bears the errors? Who can contest the decision? Massimo Ragnedda and Maria Laura Ruiu propose four questions every consequential AI audit should answer, adding a "power test" alongside the fairness test. - Tech Policy Press

dang that’s woke. but literally no one is woker than Musk if you define it as social re-engineering of reality to benefit yourself, and being absolutely triggered if anyone centers anyone other than you.