AI Reading for Friday, June 6

Google rolling out upgraded Gemini 2.5 Pro preview - 9to5Google

Google Releases Upgraded Gemini 2.5 Pro in Preview, 24-point boost in LMarena.ai , opens a stistically significant gap vs OpenAI - PYMNTS.com

Google’s upcoming Gemini Kingfall is allegedly a coding beast, appeared briefly in AI Studio - Bleeping Computer

Google rolls out experimental Gemini-powered life coaches based on real-life experts. The first of these Portraits will feature author Kim Scott, known for the book Radical Candor. - Google

Google DeepMind CEO Demis Hassabis reportedly said the company is working on a “next-generation email” tool that can reply in your style. - Gadgets 360

Google AI mode said to be rolling out dataviz for AI mode, didn't make a chart for me - Search Engine Journal

NotebookLM creators left Google after a viral hit—now they're back with their own AI app. - Business Insider

Using a clever approach, researchers find GPT-style models have a fixed memorization capacity of approximately 3.6 bits per parameter. - VentureBeat

Is China winning the AI race? - The Week

Manus has kick-started an AI agent boom in China - MIT Technology Review

China is reportedly holding up approval of Apple Intelligence as an effort to maintain leverage in the ongoing US-China trade war. - 9to5Mac

Anthropic launches new Claude service for military and intelligence use - The Verge

Ukraine shows what AGI war will look like - TIME

Defense Startup Anduril Clinches Funding at \$30.5 Billion Valuation - Bloomberg

Dual-use autonomous tractor for farming or military use (mulitary?) - The Next Web

Anysphere, whose Cursor might be the fastest growing tech product of all time, raises $900 Million - Bloomberg

AI firms say they can’t respect copyright. These researchers made an ethically sourced training set. - The Washington Post

X’s new policy prevents companies from using posts to ‘fine-tune or train’ AI models - The Verge

Meta’s shops for exclusive content on its upcoming VR headset - The Verge

Meta Oversight Board says they aren't doing enough to fight celeb deepfake scams - Engadget

Deepfake Scams Are Distorting Reality Itself - WIRED

As AI-generated content crosses the uncanny valley, the ‘zero-trust’ internet comes into view. - Fast Company

Artificial Intelligence is now an A+ law student, study finds - Reuters

PE CEO Robert F. Smith Tells Finance Conference Attendees: AI Will Take Your Job - Entrepreneur

Walmart pilots an AI-powered interview coach, a skills-matching tool and job simulator for veterans. - corporate.walmart.com

'Meth is what makes you able to do your job': AI can push you to relapse if you're struggling with addiction, study finds - Live Science

AI Diagnoses Eye Movement Disorders from Home - Neuroscience News

Artificial intelligence, dehumanisation and precarious work: translators on the frontline of tech-induced job degradation - Equal Times

Trump's AI czar, David Sacks, called federal basic income-style UBI a "fantasy." - Business Insider

He’ll be in big trouble if he tries to take it human

A DOGE staffer developed an AI tool to review Veterans Affairs contracts. But there was a slight hitch. It hallucinated the size of those deals. - ProPublica

For example, it concluded that more than a thousand contracts were each worth \$34M, when in fact some were for as little as $35,000.

Sahil Lavingia on the NYT podcast - The New York Times

9 tasks ChatGPT can handle for you in seconds, saving hours - PCWorld

A neuroscientist says it’s impossible for AI to ‘understand’ language. I kinda think if people think gradient descent on mad libs = sentience and moral rights the burden is on them to show some tests that evidence that. - The Conversation

A culture war over AI rights and obligations humans have toward AIs? - undark.org

After initially refusing to act on moderation reports, OpenAI has taken down a “Looksmaxxing GPT” that was previously featured on their GPTs page. - Business Insider

ABBA Musician Bjorn Ulvaeus Says He's Using AI to Help Write New Musical - People.com

Runway hosts AI Film Festival to spotlight movie machine intelligence - The Hollywood Reporter

Seeing SpaceX stock at, I think, > 30% discount to the last raise on private markets

From "LangGraph is trash" to "pip install langgraph": A Stockholm Syndrome StoryDiscussion - Reddit

The LangGraph state graph paradigm with human in the loop and time travel makes a lot of sense to me. They do try to be all things to all people, change everything all the time and document poorly. And I still can’t get into the discord.

Follow the latest AI headlines via SkynetAndChill.com on Bluesky