Comments

The Strait is closed, guys.

How did Mythos miss the Claude code leak?

Brad should be a permanent starter, bring on all 5 besties

“Destroying civilization and all that…” thanks Brad

David Sacks looks exhausted 🙏

The Iran war section was an amazing example of a group of billionaires who’ve been captured. Pathetic conversation and constantly turning back to the markets vs the life and death implications - they’ve become what they supposedly hated

"Apparently ... took the ceasefire to mean, level lebanon dropping 160 bombs in under 10 minutes" thank you Jcal!

I'll take Anthropic using fear as marketing any day over Apple using fear to block legislation. "If you get the camera module replaced by a third party repair shop, your mom will explode". 😂

Chris Williamson has Friedberg on as a guest with an excellent video, we almost have the tech to live forever. YouTube.

Always good to have Bard Gersner on the pod

Shrewd investors have pay attention to what they are NOT talking about in the pod. ;)

Love you David Sacks for your service .. a voice of reason. Thank you for yourr service

we need friedberg baack!! also this pod is BEST when talks AI & tech

Can we do time lapse for all of their hair lines receding?

Greg pulling a hamstring trying to talk up his stocks, reminds me of his begging SAm to keep his OpenAI shares

Sacks has been looking for any excuse to kneecap Anthropic since day one. Libertarian btw. Totally not a grudge though.

You're all complicit in this war

Bestie$ Trumpie$ Muskie$ 🎉🎉🎉

Brad has some thick skin. I go on my friends famous podcast and they all crap on my investments, most times, and yet I continue to show up. I appreciate your resilience bud.

How come this is the only channel that doesn’t have ads for non-YouTube members?

Brad's shoulder will fit in my keyboard. do some lat pull downs!

BTW the biggest danger of AI is censorship. They get to see, show,, hear, speak, learn and implement while you get none.

War ?! 1. I can't say too much. 2. The market didn't react that badly. 3. We're worried about Israel being disliked. 4. But overall, it was worth it because Iran was a threat to the United States. You're in Israeli pockets.

Tokens are the new Paper Clips

Sacks hammering Anthropic for being ethical. Isn't he the AI czar????

Change the name from the All-In Podcast to the All-Israel podcast.

nothing on Ronan's exposé?

I was hoping they'd talk about Milla Jovovich's AI play.

Damn, I did not expect that Chamath would be clearly outdated on AI. Jason just schools these fools on what is actually happening on the factory floor of AI. Chamath is an AI Boomer. I'm shocked

From here on out, it's Y2K every time a new frontier model is released..

93 of 100 senators and 361 of 435 house members take AlPAC money!

This, the All-in pod, and my own homemade cheap but tasty cup of coffee, are my much anticipated and appreciated Saturday morning treat. Thank you, gentlemen.

Chamath is right on the Anthropic.

Regarding OpenClaw, Anthropic has efficiencies built in to their products that OpenClaw does not which allows Anthropic to permit use of the generous MAX plan with Anthropic family of products. They can also prove this as Anthropic has contributed some basic optimizations for OpenClaw as a consolation for relegating OpenClaw to API use only.

Sydney, Saturday morning, sipping my long black, watching these Billy's talk candid hot topics. great way to start the weekend !

Only Brad makes sense on OpenClaw topic. It’s so evident that others like so low tech and misunderstand how those things work.

Asking Brad, an investor in Anthropic, if they made the correct decision, is like asking a tire company if you should upgrade your tires. Of course he’ll back Anthropic

I wonder how much Sacks and friends have made over the market moves after the last 5 weeks.

Many thanks! Great debates.

This week's podcast is excellent! I think Grok Auto-translate is the first of its kind. The technology as far as I know has never been used to translate communications automatically. I think it's fantastic.

Sachs is so biased that his opinion is rendered worthless. He can’t say anything that might upset daddy trump. Same with Chamath. It’s hard to watch this show sometimes.

THEY USE THE IRAN WAR TO DISTRACT FROM THE EPSTEIN INVESTIGATION, AND NOW THEY USE THE EPSTEIN INVESTIGATION TO DISTRACT FROM THE IRAN WAR. BRILLIANT, BUT OBVIOUS.

14:40 our infrastructure was already an open door hackable environment, with this they will be able to take down everything incredibly fast. Private industry sounds like they’re getting on board with trying to plug the holes but I don’t see the government moving fast enough, especially considering how they have a track record of dragging their feet on this exact issue. Edit: 22:37 EXACTLY!

The Claude agent feature is billed on API basis currently

The Anthropic “Mythos” situation feels like a stunt. In cybersecurity, rule number one is usually not to publicize vulnerabilities in a way that increases attack surface—so this raises questions. How likely is it that any partner IC was actually compromised, or that the model was truly at risk of being leaked? There’s also a lack of detail on the use cases. What environments were tested? Were these known attack vectors? And was the issue tied to model quality, or something else like extended reasoning time or system configuration? Chamat is right... these "partners" will accomplish nothing in 2 years. Plus, why this cohort selection (the criteria seems more on those who pay than the critical internet infra)? As it stands, it comes across more like positioning—creating a sense of risk and urgency—rather than a fully transparent technical disclosure.

5:30 I love how they try to make it look like a Home with someone smart with a lot of books. You should trust these guys look at all the books.

Still looking for some seeds Freidberg

What gets me is most of the money to be made on the Anthropic/ openai's of this world is happening while they are private. Seems grossly unfair.

Non-AI researchers and developers posturing to know more than the leading frontier lab is crazy

"All warfare is based on deception" ~~Sun Tzu

They looked nervous talking about AIPAC.

Jason does suck . He is dragging the show down

How is Brad invested in every single company nobody finds this bizarre?

They are terrified of criticizing the Dear Leader

Dwarkarsh Patel with Dario was an epic podcast

Brad is my favorite bestie.

There we go! What do they have to say about Mythos?!

JCal killing it! Ok I’m back to listening to this bunch 💁

Great Show! 4 different personalities that really work! With great in-depth knowledge!

5% Its a lot higher than that, we are now like 80-90% code written by AI

35 minute mark: pensioners come in to fix old code. A trillion lines of mediocre code, Gordon Knot analogy. I LOVE CHAMATH'S insights!!

Friend of the pod Jared Kushner!

All In Podcast deep pockets built in USA wealthy Ai besties, promoting and excited about sending jobs to Philippines. Geez guys, we Americans really appreciate that! That’s just what we need! Fewer jobs!

Their using fear and ignorance for marketing and operationally as a political advance The tech is so dangerous it shouldnt be so open, only a handful should be able to oversee its development (them) It is attempted moat building

"If we look at the markets today, we've bounced all the way back" "If we land the plane with Vance on these two things... I think it's off to the races" ... This is actually the sentiment I find most concerning; I fear it lacks emotional qualities founded on understanding and appreciation for the opponent (Hillbilly level understanding sir, at least we got the best of em' negotiating). Iran has endured American wars on both of it's borders for 20 years, the current negotiators betrayed them twice in 9 months, we tried bombing them, Bessent tried economic warfare while ___intelligence armed rioters, tried assassinations, threatened nuclear war... There's no emotion in your logic, and Ukraine should be a reminder of it's importance in your calculus. If you were Iran, considering pricing elasticity, supply chain logistics, the death of your president, threats of existential destruction, the fact the straight is your only leverage, you're watching Oil speculators, paper prices are completely disconnected from spot prices.... would YOU just... open up? Iran is not going to cede, they are going to throttle global economic growth until we are gone, or the Persian Gulf ceases to produce oil meaningfully (Both of which result in the same outcome).

Would be great for Brad to explain the ethics around investing in both OpenAI and Anthropic (assuming he isn't invested in Google or xAI, too). If I were Sam or Dario, that should disqualify Brad from any access to special information.

What about the ICP blockchain tech to minimize hacks?

Hacking AI Mythos is nothing compared to the power of other upcoming AI models. People have no idea that they're about to live in a world of exponentially accelerating SciFi, as reality and fiction collide. 100% chance that the world we once knew will soon be destroyed.

I recall there being claims that gpt3 was "too dangerous" to release. Were these claims concerns over safety? "play to their fear" marketing?

What eventually happens to the leading cybersecurity companies if anthropic has significantly better threat detection as it seems to?

So glad Anthropic didn’t listen to Hegseth. Respect.

Let's give credit to Jason for asking key questions on the war (that everyone else failed to answer) and apparently realizing his initial view of the war for Israel was wrong (when he interviewed the exiled "crown prince" along with fellow Uber investor Shervin) where they assumed this would result in regime change. That said, both Chamath and Brad (pro zionist) trying to downplay the impact of this war on the gulf, US and global economy was glaring and shows how biased they are (and afraid to be critical of Trump). Both Chamath and Brad would never risk their own lives or their kids lives for Israel. David Sacks is in a difficult position since he is technically associated with the admin, but obviously realizes this war for Israel is a gigantic mistake.

☝️How stupid do they think we are? OpenAI declared its language model "too dangerous to release." SEVEN YEARS AGO. Does anyone thing GPT-2 is a threat to civilization now? Anthropic is just repeating the marketing move.

translation is indeed so powerful in being able to understand each other instead of inserting our usual ignorance-based fears and prejudices. Kudos JCal for bringing it up and Chamath I expected more from you on this

Brad has Friedberg levels of thoughtfulness

Kudos to Brad for supporting holding back Mythos!

Great episode. Lots of substance/focus on business which is what we want!

JCal the D&D joke was gold

Yesssssss....my favorite intro is back 💯

friends, are we sayng that humans cannot build exploits based on multiple vulnerabilities? David's option 1 is correct = fear sales = another source of revenue. cheers gang✌🏻

Notice not one of them mentions AIPAC

10:00 their frontier model so good that it didn’t catch the bug that leaked their proprietary claude code source code in an npm package?

Great to hear All-In describe the Liquidity Event as “the new Sohn.” The Ira Sohn Conference was created in memory of my brother Ira Sohn a Wall Street trader who died at 29 years old (over 30 years ago) by his friends and colleagues as a 1 day event “Where Wall Street Comes Together to Fight Cancer,” and has since helped inspire more than 10 Sohn-modeled investment conferences globally, all supporting pediatric cancer and other related diseases. We’re proud of that legacy and would warmly invite the All-In Liquidity Event to consider donating a percentage of proceeds to the Ira Sohn Foundation to help continue this mission. 🎗📊🌍

The theme song still needs work, Jason 🦄

How do we know Mythos is not getting them to do exactly what it wants?

Man, I want to go back to punching cards.

Eli made the big show

so agree on the 'X' revolution with the 'Auto Translate' feature....kudos to Elon Musk and Nikita Bier

Isn't Sacks, the AI and crypto czar? Why does he not know about the latest happenings in the industry? It feels strange to me.

42:41 showing revenue without expenses is MISLEADING

The ending 😂😂😂😂😂😂

I think what this means is a lot of people into software are looking at this saying wow I could buy a subscription to anthropic and successfully find a radical exploit that sell for $100,000 apiece. I mean, that’s what I thought. If you think about it, its a beautiful symbiosis. Software professionals, likely recently made unemployed by AI can now purchase AI to launch their new career. It’s a win-win.

Another great debate — thank you all very much, you are truly the best!

One of the all time best episodes of the pod. Well done guys and thank you.

Thank you all for the episode.

The Anthropic glaze is crazy

Lol Chamath trying to prompt bully sacks was hilarious

The 'Ok we might be in a war, ... but the MARKETS !' was a bit of Bondi moment

Great one. Thx guys

I literally reached out to Athena while listening to last week’s pod. I had my onboarding call on Friday, and I’m supposed to be assigned an EA this week!

Brad is the best on this pod, the jcal, then science guy

Good pod!

This was a great ep

Great to have Brad to fight Chamath

Amazing Podcast!

“Flywheel” = Positive Feedback Loop (it’s Systems Dynamics baby!)

54:02 Brad shutting down smug Chamath Facebook relevance with grace is wonderful.

Great discussion

It felt like I was watching an AI-revised version of the episode from 4 weeks ago.

Thank you for looking at both sides. Good guy bad Guy is old strategy. The known "details" are swepted under the "rug" until after the big damage is done. Love this discussion, it runs deep. Think about your kids.

Well done !!

Is Washington DC wearing Sachs down? Love the guest bestie. Thanks

"Well maybe they are customers of them" lol

You guyss are awesome I love the show I wish I could befriends with you all in real life. Keep up the great work!

Thank you.

Minute 23, Chamath's point about airplanes is BRILLIANT. Chamath has been my favorite voice in this episode. Followed by the quip that "maybe airplane companies are the customers of the (companies Dario went to)" is a good quip but I hope to friggin God Dario heard Chamath and is now calling anyone doing fission and big mass times big velocity (airplanes+)

Give Jason hell, we will just love him more!

What is the Tao project Jason mentioned

38:27 linkedin tried to do something like this a couple years ago- asking experts to “contribute”

Just to be clear moonshots is the number podcast in the 🌎

Great pod! Would love to hear a discussion concerning Elon and Intel

The open was critcile !!! so nice to see David Back !!!

brad g is awesome

i enjoyed the tech talk, thats when this show is at its best. getting a nobel peace prize for auto translate is hilarious though :DDD

Good pod

The best podcast is when none of the besties r candid n don't have ' investments ' in the topics being discussed. 😂 This was one of them. None, except Brad have a stake in Anthropic. Brad would be wise to reconsider whether his marketing is backfiring on his investments 😄

would love to hear a take on the burning China hack

I saw a very expensive green sports car driving through SW Minnesota with the vanity plate that read All In I tried to keep up but at about 90mph I had to back off. I wanted to see make of car and the state.

New intro is pretty!

As this AI journey rolls on, will we humans be happier and more fulfilled?

At the start i read it as Anthropic’s $30B Ram, which is also true probably.

Brad on some serious cool aid

No matter how powerful a new model is - it's most dangeorus while being used by bad actors for bad use cases from the core where it spawned or created. But once it get released to the wild, the balancing do kick-in from the edge. The hope is that bad actors never get sole and full access to the core that is not balanced by the edge. A frontier model is most dangerous when its core becomes accessible to bad actors while the corrective intelligence of the edge is absent, delayed, or structurally weakened. Release into the wild does not automatically make it safe, but it can activate balancing dynamics that closed power suppresses. GenAI is hopeful because the edge matters, but existentially risky because if madness captures the core that governs access and direction, the edge may inherit a poisoned field too late to rebalance it.

What makes you guys think it’s a 1 time thing? I also never heard the argument; let it run wild now and take the damage, that way people take it seriously now, rather than waiting for down the road when someone figures out how to build a nuke

"It's like Avengers assemble against Ultron"

1:15 keep fighting the good fight Jason humiliate them in the openings never give up

They don’t seem to consider that the release of these warnings could have steered other companies to adjust these models to not behave in the ways they were concerned about.

'Nothingburger' was on my Bingo card! Thanks Chamath lol Is what Dorsey is doing at Block a nothingburger? Even the way they have gone about layoffs is worth a discussion IMO. Then his blog post which attempts to rewrite organizational structure and design is intriguing. I feel like everything that they are doing could carry a full episode. Thoughts??

Thanks Chamath for saying the truth about the reality of enterprise deployments. Please send Jason to do an internship for a few weeks at a MNC IT department to understand the revolution moves at a different speed there.

Picking bedsheets can be tricky. Which fabric is right for your sleep style? What's the best thread count, and does weave type matter? Before you buy your next set of sheets, find the answers to all these questions.

25:42 they did a variation of this a month or two ago.

I can’t make out the name of the guy they are talking about in the first minutes, which Chamath says Marc Andresen found? Anyone who can type out the actual name or maybe provide some links?

I don't think it's the models that are the big risk for security. I think it's more the harness. ie. opencode, or claude code. It's very hard for a person to check for 50 well know security vulnerabilities, but it's relatively easier for opencode with any model

Jcal gets so nervous whenever sacks starts talking.

Folks please release subsequent content from guest speakers or interviews at the liquidity event. Leave something for us mids

LLMs and SLMs are pure commodities, the orchestration and ontology layer of PLTR is crucial to the enterprise...

We play every Friday!

Disagree with so much of the political analysis here but mad respect and appreciate that they don’t run ads

Brad is the most sensible on this panel. He brings an objective and analytical mindset to the borderline disorganized, chaotic, and biased rambling of few members. ESP the govt guy, whose full-time job it seems is just to shit on companies which he doesn't like and try to play bigger than thou.

@chamath you crack me up, look at his prompt!!!

Only Jay Cal could complicate sliced bread 🍞

Grok - whats that lol

For the ones wearing In Ear Monitors what In Ear monitors is everyone using in the All-In Pod?

Is Brad Gerstner the inspiration for the character Russ Hanneman on the TV show Silicon Valley?

Time for a new Intro/Themesong

1 million dollars is the cost of two software maintenance consultants.

Markets aren’t reacting to headlines anymore—they’re pricing duration, not shock. The longer this drags, the more pressure on liquidity.

Anthropic uses fear, but Chamath doesn’t when he says the internet needs to be shut down for 5 years.

Where can I have Amazon send a desk pad for Chamath’s desk to stop the acoustic torture that his desk pounding creates?

I wouldn’t say those vulns were “missed” Bob, ;)

Friedberg present last week - no cold open. Friedberg not present this week - cold open check. I'm drawing my own damn conclusions

Curious that Jason describes the Iran war as the “3rd rail?”

That ending though!!!! 😂😂😂 Jason is hilarious. I don’t agree with most of his political views but dang he can be funny

Brad playing elite defense for Anthropic 😅 33:09

30:40 the people are with Jcal on that

Anthropic is billing big clients millions, but they don't have to pay. They put the enterprises on the API usage, but the they don't have to pay. This is happening even in relatively small banks in Australia.

Work hard! Don't let my browser history leak! If you could show me Chamath's browser history, that would be perfect!

you know how heavy their Anthropic bags are when they spend entire episodes talking about AI and shitting on OpenAI

34:46 saying this so early in the game. INSANE.

@35:00 When @chamath lights up the podcast with reality, I can say as someone on the other side experiencing a 30 year tech debt data science project myself, it's like looking into the Abyss. NO BOTTOM.

Chamath you are wrong at 35:14 claude opus is able to code legacy languages really well. I am a software engineer who has never coded COBOL before and recently was assigned to a team that works mainly in COBOL. Claude has been doing all of my work almost perfectly