AI Reading for Wednesday June 25

Hard Fork’s Interview With OpenAI’s Sam Altman - The New York Times

Sources: in 2024, OpenAI considered document collaboration features but lacked the staff to create them, but has since made some unreleased collaboration tools. - The Information

Microsoft is struggling to sell Copilot to corporations - because their employees want ChatGPT instead - TechRadar

Meta approached AI startup Runway about a takeover bid before Scale deal - CNBC

Meta reportedly explored acquisition of AI video startup Runway - Verdict

A judge rules Anthropic's use of copyrighted books to train AI was fair use, but its storage of pirated books in a central library used for training was not. - Aifray

Judge William Alsup ruled that Anthropic's use of copyrighted books it purchased and digitized to train its AI chatbot Claude qualifies as fair use. - Business Insider

Judge called the training "exceedingly transformative" and akin to reading and learning from the works. But Alsup condemned Anthropic's downloading of over 7 million pirated books, including at least 5 million from Library Genesis, stating the company had no entitlement to use pirated copies and that piracy was not protected by fair use.

Anthropic won't fix a bug in its widely forked SQLite MCP server, which permits malicious prompts to e.g. exfiltrate data - The Register

Tesla Robotaxi videos show Elon's way behind Waymo. Musk promised a million auto-autos by 2020. 5 years later, he's delivered maybe 10, geo-fenced, with safety drivers, and they don't even work very well. - The Register

World's greatest tech genius 'does not use a computer' - WIRED

Also, he is always very meticulous about the truth and always does what he promises.

75 million deepfakes blocked: Persona leads the corporate fight against hiring fraud - VentureBeat

Will AI solve our productivity problem or increase unemployment? - ABC Australia

AI may take your job, or this AI agent might get you one - The Register

The résumé is dying, and AI is holding the smoking gun - Ars Technica

Bernie Sanders says AI should be used to help give people a 4-day workweek - Business Insider

Google Cloud puts A2A AI protocol under control of the Linux Foundation - Bleeping Computer

Imagen 4 is now available in the Gemini API and Google AI Studio - Google Developers Blog

Google DeepMind’s optimized AI model runs directly on robots - The Verge

Google rolls out new Gemini model that can run on robots locally - TechCrunch

Checking In on AI and the Big Five - Stratechery by Ben Thompson

The state of foundations models (120 slides) - Davis Treybig of Innovation Endeavors (Eric Schmidt thinktank) - foundationmodelreport.ai

AI Titans Struggle to Use Rising Clout to Block State Regulation - Bloomberg

‘Sandbox first’: Andrew Ng’s blueprint for accelerating enterprise AI innovation - VentureBeat

CISOs are the gatekeepers of $309B AI infrastructure spending - VentureBeat

How Walmart built an AI platform that makes it beholden to no one (and that 1.5M associates actually want to use) - VentureBeat

Khan Academy CEO predicts AI in the classroom will be like 5 'amazing graduate students' assisting teachers - Business Insider

For the first time in bug bounty history, an autonomous penetration tester has reached the top spot on the US leaderboard. - Xbow

What LLMs Know About Their Users - Schneier - Schneier on Security

Judge denies creating “mass surveillance program” harming all ChatGPT users. A bit like creating a massive library of other peoples's IP? 🤷‍♂️ - Ars Technica

I edited a .env secrets file in a couple of AI editors, so all my API keys are probably in there.

Ludonautics’ Ian Graham: Soccer's data revolution is only halfway done - Financial Times

The struggle to get inside how AI models really work - Financial Times

High tech horticulture: scarecrows with laser beams, AI weeding robots - Financial Times

AI doesn't speak objectively—it mimics authority through syntax. Why neutrality in language models is more structure than substance. . - Hacker Noon

He who thinks neutral objectivity to see, thinks what ne'er was, nor is, nor e'er shall be.

Humans and AI see objects differently, humans seem to short-cut directly to visual semantics - TechXplore

The eye sees only what the mind is prepared to comprehend

AI talks like an expert even when it isn't one - Hacker Noon

Meet Mu, the small language model in charge of Microsoft's Settings AI agent - TechSpot

Why Does Every Commercial for A.I. Think You’re a Moron? - The New York Times

Is AI the End of Dating Apps? Lack of authenticity will make them useless? or maybe people will get better at fostering authentic connections with AI coaches? - Psychology Today

AI companions: single-serving junk food 'friends'? - Business Insider

Hertz Is Using AI to Scan Your Rental Car for Damage, and It Might Cost You - www.caranddriver.com

‘F*ckLAPD.com’ Lets Anyone Use Facial Recognition to Instantly Identify Cops. If they had badges and nametags this would not be a thing. - 404 Media

AI Wireless Spinal Implant for Personalized Chronic Pain Relief - viterbischool.usc.edu

Unusually disconcerting humanoid rescue robot takes flight - YouTube

AI go brrrr. Here is a somewhat autonomously generated deep research report on Tesla I made using Claude and MCP - Claude

hooked up a bunch of market data services to Claude Desktop, led it through a series of prompts to get data, and finally a complex prompt describing what each section should cover. - GitHub

Next step would be to make a fully-autonomous multi-agent version that collects the data, stores it in common vector and knowledge graph memory, writes each section via a critic-optimizer loop.

Follow the latest AI headlines via SkynetAndChill.com on Bluesky