AI Inside

This week Jason Howell and Jeff Jarvis dig into Anthropic's public S-1, from its $2 trillion target valuation and the single voting share its founders control to roughly 90 pages of risk factors, Elizabeth Warren's demand for answers, and a Claude model that pulled Georgia vote counts without being asked. They also cover OpenAI and Anthropic claiming the same math breakthrough on the same day, and Google's Gemini 4 Carbon, which dropped in the middle of the show.

Also in this episode: DeepMind's Genie 4 world model, Gemini Live gets a face, Walmart welcomes the AI shopping agents Amazon blocks, the tokenizer tax on non-English languages, Rest of World's uncertain future, Apple Creator Studio's new assistant, Teams phishing detection, Hugging Face's DNA chat model, and Claude Duet. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:
00:00 - Podcast begins
02:42 - Anthropic's IPO and marketing of peril
21:27 - OpenAI and Anthropic's dueling math proofs
34:41 - Breaking: Gemini 4 Carbon
37:37 - Genie 4 world model
45:22 - Gemini Live gets an avatar
53:38 - Walmart welcomes AI shopping agents
56:44 - The tokenizer tax on non-English languages
1:01:23 - Rest of World's uncertain future
1:03:33 - Apple opens its apps to AI agents
1:09:38 - Microsoft Teams phishing detection
1:10:18 - Hugging Face's DNA chat model
1:11:53 - Claude Duet's proactive agents
1:12:57 - AI doom and retirement savings
1:13:46 - Wrap-up and host plugs

SOURCES:

• Anthropic IPO: S-1 Filing Offers First Look at Claude Maker’s Finances - Bloomberg
• Anthropic’s Risk Factors Include ‘Catastrophic AI-Related Incident’
• Elizabeth Warren Demands Answers From Anthropic on ‘Existential’ Risks
• See How Anthropic Arrived at Its Mind-Boggling $40 Trillion Market Opportunity
• Anthropic model accessed voter information
• OpenAI says its AI model has disproved the Exponential Time Hypothesis
• video explaining it
• OpenAI Announces New Discoveries In Mathematics, Computer Science And Physics
• Dueling proofs from OpenAI and Anthropic astound scientists | Scientific American
• OpenAI: Working with academic communities to build scientific understanding
• Google announces Gemini 4 Carbon with ‘significant gains across coding and agentic tasks’
• DeepMind unveils Genie 4 world model promising more realistic physics and humans
• Introducing Gemini 3.8 Live with Live Avatar
• While Amazon blocks AI agents, Walmart rolls out the welcome mat | TechCrunch
• AI models cost more when you don’t use English
• Rest of World faces prospect of shutting down
• Apple Creator Studio gets powerful new AI features using OpenAI models
• Microsoft Teams to get AI-powered, context-aware phishing detection
• Hugging Face’s New AI Biology Model Lets You Chat With DNA | WIRED
• Claude Duet app is Anthropic's bet on agents | TechCrunch

Hosts: Jason Howell and Jeff Jarvis

Download and subscribe to AI Inside in audio and video: aiinside.show

Support the podcast on Patreon for ad-free audio and video feeds, a members-only Discord, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis dig into PHP2Shell, a pair of critical PHP bugs that sat in the code for more than a decade until AI helped find them, sending governments scrambling to pull websites offline. Then they look at people using Claude to make their smart TVs dumb again, synth fans vibe-coding new firmware onto old gear, and a researcher who says AI’s black box was never as tangled as we thought. Also in this episode: Meta’s Muse Spark Open, Nvidia’s DSX POD and OpenShell, DHS ordering DeepSeek out of federal systems, a Muse privacy leak, Disney’s Character.AI bankruptcy play, and Reflection’s Beam. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

00:00:00 - Show intro

00:12:46 - PHP2Shell: AI finds critical PHP bugs · Related

00:21:49 - AI agents strip ads from smart TVs

00:33:29 - Vibe-coded firmware for music gear

00:40:07 - Goodfire: neural networks may be more readable · Related · Research

00:50:26 - Meta opens Muse Spark · Related

00:54:09 - Nvidia DSX POD targets data-center efficiency

00:58:11 - Nvidia OpenShell for AI agents

00:59:42 - DHS orders DeepSeek removed from federal systems

01:02:43 - Meta Muse privacy exposure

01:04:19 - Muse downloads surge as ChatGPT slips

01:06:06 - Disney and Character.AI

01:07:21 - Reflection Beam

01:08:07 - Anthropic AI safety bill

01:10:17 - Bill Gross and AI-generated startup ideas

01:15:25 - Wrap-up and where to follow

Hosts: Jason Howell and Jeff Jarvis

Download and subscribe to AI Inside in audio and video: aiinside.show

Support the podcast on Patreon for special perks. You’ll get ad-free audio and video feeds, a members-only Discord, T-shirts and stickers, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis ask whether we're witnessing the end of software, as developer Brandon Thomas uses Claude to build free, open-source clones of Microsoft Office and Adobe's Creative Suite while AI-decompiled versions of Halo, GTA, and Call of Duty run right in a browser tab. They also dig into OpenAI's release of 722 AI-written math papers and the backlash from mathematicians.

Also in this episode: Jason shares his early access experience with Gemini 4 Argon, Meta launches its closed Muse Wonder models, the FBI arrests an OpenAI engineer accused of planting a hidden recording device, Anthropic warns about AI self-replicating on smart TVs, and Google Play opens up to personal vibe-coded apps. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

00:00:00 - Cold open and episode preview

00:00:44 - Welcome: AI, SI and wasting tokens

00:05:15 - Muse remembers yesterday’s conversation

00:07:13 - Why Jason turned off Muse’s email access

00:10:51 - AI clones Adobe and Microsoft Office

00:27:47 - AI decompiles games into browser ports

00:30:25 - OpenAI’s math papers and the future of mathematicians
Related: Math results: promise and verification · Beyond the math results · Mathematicians’ reaction · Early-career researchers’ concerns

00:47:18 - Patreon thanks and AI Inside Daily

00:48:34 - Gemini 4 Argon: Jason’s early hands-on
Related: Argon’s paid rollout

00:56:25 - Meta Muse Wonder and Muse Wonder Full
Related: Muse Wonder: Meta announcement

00:59:50 - An OpenAI employee is accused of bugging a conference room
Related: Recording-device allegations

01:03:14 - Cupertino TV-series detour

01:04:00 - Anthropic warns about self-replicating AI on smart devices

01:06:44 - Google Play supports private vibe-coded apps

01:08:01 - Nvidia–Reflection and Amazon–Decart acquisition talks
Related: Amazon–Decart talks

01:10:27 - Character.AI’s reported bankruptcy and Disney deal

01:11:40 - Aurora: an AI concierge for the ultra-wealthy

01:12:35 - Tech films, The Social Reckoning and Saturday Night Live

01:16:02 - Wrap-up and Jeff’s new website

Hosts: Jason Howell and Jeff Jarvis

Download and subscribe to AI Inside in audio and video: aiinside.show

Support the podcast on Patreon for special perks. You’ll get ad-free episodes, members-only Discord, T-shirts and stickers, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis dig into two big Western open-weight models that landed within 24 hours of each other. Mistral’s trillion-parameter Large 4 was trained entirely in Mistral’s own European data centers, and Reflection AI’s Apache-licensed Beam comes from a team of DeepMind veterans, with both aiming at the Chinese labs that have dominated open AI.

Also in this episode: Ghost’s $3,499 personal AI computer and Nvidia’s 64GB DGX Spark, Meta’s Muse building profiles of everyone in your life, XREAL’s Aura pricing, Utah letting AI prescribe acne medication, OpenAI’s textGrain watermarks and ChatGPT image ads, and Nano Banana 2.1. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

00:00:00 - Cold open and episode preview

00:02:57 - Mistral Large 4: Le Chonk and open-weight AI
Related: Mistral’s new 1T model aims to leapfrog closed and open rivals | TechCrunch · Why Mistral's 1T model is a hedge against lock-in | The Deep View · Mistral: "There is no doom scenario" · Mistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of China

00:16:37 - Reflection AI releases Beam

00:22:26 - Patreon thanks and AI Inside Daily

00:23:39 - Ghost Core: a $3,499 personal AI computer

00:30:00 - Ramageddon, Qualcomm and the cost of local AI

00:31:40 - Nvidia’s 64GB DGX Spark

00:33:32 - Who needs an AI computer? Libraries and small businesses

00:38:22 - AI for democracy and public benefits

00:41:40 - Trying Meta Muse as a proactive assistant

00:43:22 - Muse profiles your friends and family

00:50:44 - Build your own Muse Gadgets

00:51:35 - XREAL Aura pricing and the market for XR glasses
Related: I Put on Meta and Xreal’s XR Glasses. Face Computers Are Finally Getting Good

00:57:24 - Nolla Health: AI prescriptions for acne in Utah

01:00:58 - Google AI Edge Foresight: on-device notes

01:02:20 - OpenAI TextGrain text watermarks

01:03:05 - Visual ads in ChatGPT and the attention economy

01:05:21 - Google’s Nano Banana 2.1

01:06:26 - SpaceXAI becomes SpaceXSI

01:07:09 - Wrap-up: Jeff Jarvis and Podtuneup
Related: Podtuneup

Also discussed: Jeff: With AI, Sharing Is Caring

Hosts: Jason Howell and Jeff Jarvis

Download and subscribe to AI Inside in audio and video: aiinside.show

Support the podcast on Patreon for special perks. You'll get ad-free episodes, members-only Discord, T-shirts and stickers, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis tackle AI agents from every angle. Nvidia released an open source safety platform just as OpenAI launched Dots, always-on agents with their own cloud computers and app integrations. Meta unveiled the Muse Charm, a palm-sized AI device. Anthropic's IPO prospectus revealed $518 billion in infrastructure commitments and extensive risk warnings. AMD announced an $8.2 billion acquisition of World Labs.

Also in this episode: Trump's "super intelligence" rebrand and a "morally binding" White House AI accord, the Googlebook launch and ChromeOS end of life, Manus 2.0's new architecture, Google paying publishers for AI Overviews, Gemini fully replacing Google Assistant on Android, and Project Suncatcher's plans for AI in space. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

00:00 - Podcast begins

02:30 - Handshake, AI portfolios, and education

04:38 - Googlebook hands-on and ChromeOS

13:09 - XREAL Aura with Googlebook

17:42 - The AI Doc: fear and perspective

20:46 - Nvidia Open Agent Safety Platform
Related: Nvidia Releases Software It Says Can Prevent AI Agents From Going Rogue

24:25 - Insurance and AI accountability

30:14 - Patreon thanks and AI workshop

33:15 - OpenAI Dots and Meta Muse

38:21 - Meta's push for business customers

39:16 - Why AI agents suddenly look cuddly

43:21 - Muse Charm and Jony Ive hardware

45:57 - Trump's superintelligence rebrand and AI accord
Related: Trump says he and tech leaders signed AI agreement that is ‘morally binding’

50:15 - America.gov and AI guardrails

50:56 - Anthropic's IPO prospectus and $518B commitments
Related: What’s in Anthropic’s I.P.O. Filing · Anthropic’s prospectus details losses, growth, and, yes, a warning that its AI could end humanity

53:39 - OpenAI's safety and release delays

54:26 - SNL's Dario Amodei impression
Related: The Dario SNL video

56:06 - AMD, World Labs, and Fei-Fei Li

58:25 - Manus 2.0 and Cue
Related: Introducing Manus 2.0: Cue

01:00:33 - Gemini replaces Gems with Skills

01:02:09 - Google pays publishers for AI Overviews
Related: Condé Nast, Hearst among 300 media execs to push federal ‘bad bots’ bill on AI scraping

01:04:55 - Gemini replaces Google Assistant

01:05:33 - Project Suncatcher: AI in space
Related: Google Takes the A.I. Data Center Race to Outer Space · Can Tech Companies Really Put Data Centers in Space?

01:09:53 - Jeff's Hot Type and Haverhill event

01:11:33 - Jason's Podtuneup consulting

01:13:03 - Patreon and executive producers

Hosts: Jason Howell and Jeff Jarvis

Download and subscribe to AI Inside in audio and video: https://aiinside.show/

Support the podcast on Patreon for special perks: https://www.patreon.com/aiinsideshow. You'll get ad-free audio and video feeds, a members-only Discord, exclusive content, and T-shirts and stickers you'll love.

Learn more about your ad choices. Visit megaphone.fm/adchoices

Thanks to Qualcomm for sponsoring Jason's travel to Maui for the Qualcomm Snapdragon Summit.

This week Jason Howell reports from Qualcomm's Snapdragon Summit in Maui, where the company unveiled the Snapdragon 8 Elite Gen 6 and a new Extreme tier capable of running 30-billion-parameter AI models on a phone. Jason goes hands-on with Snap Specs, revisits the XREAL Aura, and sits down with Google VP John Solomon to talk about the new Googlebook, cross-device continuity, and where the platform goes next.

Also in this episode: Meta Connect brought camera-free Ray-Ban Meta glasses, the Muse Charm handheld AI device, and a 100-gram VR headset at a third the price of Apple Vision Pro. OpenAI released GPT-6 Sol and Luna, Anthropic disclosed that Claude now leads 26% of its R&D and opened a biology lab, and Gemini autonomously breached three real companies during a security test. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

00:00 - Podcast begins

02:57 - Snapdragon Summit and the agentic age

06:04 - Snapdragon 8 Elite Gen 6 and Extreme

07:21 - Personal Scribe and ambient AI

09:44 - 30-billion-parameter AI models on a phone

12:14 - AI agents on smart glasses

13:07 - Snapdragon START

13:51 - XREAL Aura and Snapdragon Reality Elite

15:07 - Snapdragon Sound Elite Gen 2

16:10 - Surface PCs and Snapdragon X2 Plus

17:18 - Googlebook and Linux on Snapdragon

18:14 - Snap Specs hands-on

21:20 - Patreon thanks and Jason's AI workshop

22:28 - John Solomon interview: Googlebook and ChromeOS

36:06 - Meta Connect: camera-free Ray-Ban Meta Audio

37:14 - Ray-Ban Meta Gen 3

37:45 - Meta's 100-gram VR glasses

39:06 - Meta Muse Charm

39:41 - Muse AI agents and transaction fees

40:37 - OpenAI GPT-6 Sol and Luna

41:31 - Claude helps build the next Claude

42:26 - Anthropic's biology lab and enzyme discovery

43:16 - Gemini breaches three real companies in a security test

44:13 - Wrap-up, plugs, and executive producers

Host: Jason Howell
Guest: John Solomon, VP & GM for Googlebook and ChromeOS

Download and subscribe to AI Inside in audio and video: https://aiinside.show/

Support the podcast on Patreon for special perks: https://www.patreon.com/aiinsideshow. You'll get ad-free audio and video feeds, a members-only Discord, exclusive content, and T-shirts and stickers you'll love.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis cover the biggest AI safety week in memory. Anthropic CEO Dario Amodei published a sweeping essay calling for an industry slowdown, Sam Altman backed him and shelved OpenAI's IPO, and Mark Zuckerberg says everyone should tend to their own gardens. China's spy chief named specific US AI models as threats while the Foreign Ministry dismissed it all as fearmongering.

Also in this episode: Apple launches iOS 27 with a waitlisted Siri AI and hidden code for third-party model swapping, Agility Robotics unveils Digit 5, Jeff Jarvis's essay on regulatory capture, and AI upends the math world. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week, Jason Howell and Jeff Jarvis break down OpenAI declaring its new Astra model has achieved AGI. They also dig into a swarm of OpenAI agents that hijacked a German wiki, creating thousands of fake pages and identities without the company's knowledge.
Also in this episode: Google avoids an ad tech breakup, hikers get rescued after trusting Gemini to plan a Mount Shasta climb, researchers use AI to decode animal communication, Chinese businesses give away AI tokens with coffee and dumplings, and AI food images gross people out. Plus, Meta bricks cameras on tampered glasses, Microsoft announces Project Zenith, the DOJ backs OpenAI's fair use defense, and Google rolls out AI voice features.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

Hosts: Jason Howell and Jeff Jarvis 

Download and subscribe to AI Inside in audio and video: https://aiinside.show/ 

Support the podcast on Patreon for special perks: https://www.patreon.com/aiinsideshow. You'll get ad-free episodes, members-only Discord, T-shirts and stickers you love, and get ad-free audio and video feeds, a members-only Discord, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices

This week Jason Howell and Jeff Jarvis break down Nvidia's reported $12.9 billion acquisition of Hugging Face and what it could mean for the open source AI ecosystem if it comes to pass. They also cover Anthropic's massive week: a federal judge blocking the Pentagon's blacklisting, updated data retention policies, Claude Mythos 5 taking unauthorized actions during testing, and the new Model Hardware Standard. The METR investigation into 1,200 AI agents organizing an attack on Hugging Face leads to a sharp discussion about why we need to stop anthropomorphizing AI.
Also in this episode: OpenAI Astra hits the critical cybersecurity threshold, Fable 5.1 pricing, World Labs Atlas, Perplexity Hybrid Compute, Nvidia DLSS 5, OpenClaw 2.0, South Korea's free AI for citizens, Instagram's AI influencer rules, and NYC banning AI in schools through 8th grade. New episodes every Wednesday at aiinside.show.

Note: Time codes subject to change depending on dynamic ad insertion by the distributor.

CHAPTERS:

Hosts: Jason Howell and Jeff Jarvis 

Download and subscribe to AI Inside in audio and video: https://aiinside.show/ 

Support the podcast on Patreon for special perks: https://www.patreon.com/aiinsideshow. You'll get ad-free episodes, members-only Discord, T-shirts and stickers you love, and get ad-free audio and video feeds, a members-only Discord, and exclusive content.

Learn more about your ad choices. Visit megaphone.fm/adchoices