Unchained

Anthropic IPO pricing, AI cybersecurity shifts, and DeFi security evolution discussed on Unchained

1h 6m Listen to the original
Anthropic IPOAI ModelsCybersecurityDeFi SecurityGoogle AIMeta AIOpenAIRed Teaming

Summary

Kain Warwick details Anthropic's IPO pricing at $70-$78 per share for a $1.4 trillion valuation, characterizing the investment as existentially risky due to the company's aggressive AI development. Austin Griffith contrasts this with Google and Meta's successful AI integration, arguing that Anthropic's internal governance and organizational structure pose greater execution risks than external competitors. The hosts discuss the competitive landscape of AI models, noting Anthropic's Opus 5.5 dominance against OpenAI and Meta. Later segments cover the Optimizely injection, the role of AI in cybersecurity attacks versus defense, and the evolution of DeFi security from traditional audits to AI-driven circuit breakers and continuous red teaming.

Episode sections

  1. 1:17 – 6:49

    Anthropic IPO Details and Existential Risk

    Kain Warwick outlines Anthropic's IPO pricing at $70-$78 per share for 825 million Class A shares, raising $64 billion at a $1.4 trillion valuation, with up to 20% allocated to retail. He characterizes the investment as existentially risky, suggesting the company is actively pursuing capabilities that could threaten humanity, unlike previous tech giants.

  2. 6:49 – 14:31

    AI Model Landscape and Anthropic's Competitive Position

    Warwick notes that while OpenAI's Astra 6.2 is strong, Anthropic's Opus 5.5 remains dominant and preferred by users, providing Anthropic with a competitive lead ahead of its IPO. He contrasts this with Meta and Google, who have made significant leaps but are still perceived as playing catch-up in the state-of-the-art race.

  3. 14:31 – 19:40

    Governance and Organizational Risks

    Austin Griffith identifies Anthropic's internal governance, organizational structure, and culture as the primary investment risk, rather than existential AI threats. He argues that while Google and Meta face challenges in cannibalizing legacy revenue, Anthropic's 'weird org structure' and moral grandstanding pose a greater threat to execution.

  4. 19:40 – 27:15

    Google and Meta's AI Integration and Monetization

    Griffith explains that Google and Meta are successfully integrating AI into their core products, with Google's AI search results becoming a primary user interaction point. The hosts discuss the shift from free search to subscription models or ad-supported AI, noting that Google's scale allows it to dominate the 'plumber beam' of practical, everyday queries.

  5. 27:15 – 33:25

    Anthropic IPO and Token Exposure

    Speaker_1 discusses the Anthropic IPO, comparing it to the SpaceX IPO and arguing that exposure to AI companies is necessary for those invested in crypto, citing a quote about not wanting to be exposed to the industry one likes.

  6. 33:25 – 40:02

    Optimizely Injection and Low Theft

    The hosts discuss the Optimizely injection affecting sites like Ledger and HSBC, noting the surprisingly low theft amount of $7,000 despite the broad impact, which Speaker_3 attributes to unskilled attackers.

  7. 40:02 – 42:34

    AI in Cybersecurity and Human Code Vulnerabilities

    Speaker_0 argues that current hacks are due to human-written code being exploited by AI, not AI writing the code, and discusses the transition period where AI is better at attacking than building.

  8. 42:34 – 44:34

    DPRK AI Usage and Future Threats

    Speaker_3 notes that DPRK's use of AI is currently limited to optimizing social engineering scripts, but predicts that AI-native attackers will eventually create AI-first campaigns.

  9. 44:34 – 46:20

    Normie Perception vs. AI Capability

    Speaker_0 highlights the disconnect where 'normies' fear AI hacking but refuse to let AI write critical software like flight tracking or surgery tools, despite AI potentially being safer.

  10. 46:20 – 47:13

    Builder Enablement and Red Teaming

    Speaker_1 asks Speaker_0 about the importance of red teaming and formal verification for builders, referencing the effectiveness of 'ETH skills' and Claude in security testing.

  11. 47:13 – 50:59

    AI Audits vs Traditional Audits

    Speaker_0 argues that while AI audits (like Mythos) are becoming cheaper and more integrated into CI/CD pipelines, they are not yet on par with expensive human audits. Speaker_1 notes that human audit firms are inconsistent due to crypto market cycles, whereas AI offers a more stable, albeit currently less effective, alternative for catching edge cases.

  12. 50:59 – 57:00

    Continuous Red Teaming and Monitoring

    Speaker_1 describes running continuous red team exercises using AI agents to simulate attacks on their own accounts, viewing these attempts as opportunities to test defenses. Speaker_3 advises crypto teams to prioritize monitoring and detection over traditional audits, as AI agents can quickly identify anomalies and low-brow attacks that humans might miss.

  13. 57:00 – 1:02:12

    Circuit Breakers and Risk Mitigation

    Speaker_3 proposes a layered security approach where audits cover the majority of risks, but AI-driven circuit breakers handle the long-tail 10% of incidents. This involves defining strict limits (e.g., no more than $200 million leaving a contract) and automating responses to prevent total loss, shifting the focus from preventing all hacks to mitigating catastrophic failure.

  14. 1:02:12 – 1:06:16

    The Evolution of Kill Switches in DeFi

    Speaker_1 contrasts the historical resistance to external kill switches with the modern acceptance of immutable, code-based circuit breakers. They argue that having a system that can pause operations (like Aave) is preferable to an immutable contract that results in a total loss, citing the DAO hack as an example where time-locks allowed for a White Hat recovery.

Assets discussed

AssetKindStanceContext
NVIDIANVDAequityneutralUm, so uh so it's been a crazy week. Uh NVIDIA lost its CEO. Maybe he comes back.
AlphabetGOOGLequityneutralLike there was a vague sense um that Google could. even though they were saying, like, don't be evil and all of that stuff, right? Like, there was a
MicrosoftMSFTequityneutralwe had Microsoft go down.
AppleAAPLequityneutralthe scariest one, genuinely to me, the scariest one, because this happens so infrequently, um, was Apple.
BitcoinBTCcryptoneutralThey do a uh send us one Bitcoin and we'll send you two Bitcoin back campaign.
Ethereum ClassicETCcryptoneutralEthereum Classic. Fucking thanks, Mary. Um, but yeah, so uh, so I just I think that like the design space of ensuring that we've got like systems that can, you know, have kill switches and stuff like we've sort of accepted at this point that that is um far preferable to an immutable smart contract uh that you know has no escape hatch at

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Anthropic is raising $64 billion in cash at a $1.4 trillion valuation in its IPO.

Unidentified speaker (UNKNOWN) ·3:37

How to verify: Check the S-1 filing with the SEC EDGAR for Anthropic's IPO registration statement for the offering size and valuation metrics. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic is selling 825 million Class A shares between $70 to $78.

Unidentified speaker (UNKNOWN) ·3:31

How to verify: Verify against the price range and share count listed in the final prospectus filed on SEC EDGAR. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Up to 20% of Anthropic's IPO shares are set aside for retail investors.

Unidentified speaker (UNKNOWN) ·4:18

How to verify: Review the 'Plan of Distribution' section in the S-1 filing on SEC EDGAR. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Opus 5.5 has been dominating as the preferred model for six weeks prior to the IPO.

Unidentified speaker (speaker_1) ·9:44

How to verify: Cross-reference with AI benchmark leaderboards (e.g., LMSYS Chatbot Arena) and user sentiment reports from the period leading up to the IPO. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

OpenAI released Astra 6.2, which is good but not a threat to Anthropic's lead.

Unidentified speaker (speaker_1) ·9:04

How to verify: Check OpenAI's official blog or press releases for the release date and details of Astra 6.2. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Meta released Muse Wonder Wonderful Full 1.0, a powerful but 'spiky' model.

Unidentified speaker (speaker_1) ·11:39

How to verify: Verify via Meta AI's official announcements or technical blog posts regarding the release of this model. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Google has made a 'massive step change' in AI, leading to increased user subscriptions.

Unidentified speaker (speaker_1) ·13:02

How to verify: Check Google's financial reports for changes in AI-related revenue or subscription metrics, and user adoption data from third-party analytics. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Anthropic's internal governance and organizational structure are the biggest risks to its success.

Unidentified speaker (speaker_1) ·14:31

How to verify: Review the 'Risk Factors' section of Anthropic's S-1 filing on SEC EDGAR for disclosures on governance and organizational risks. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Google's AI search results are becoming a primary user interaction point, replacing traditional browsing.

Unidentified speaker (speaker_3) ·21:19

How to verify: Analyze user behavior data from Google or third-party search analytics firms showing shifts in click-through rates and AI result interactions. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Apple is 'asleep at the wheel' in AI compared to Google and Meta.

Unidentified speaker (UNKNOWN) ·16:26

How to verify: Compare Apple's AI product releases and market share in AI services against Google and Meta using industry reports and financial filings. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic is raising $81.3 billion at a $1.5 trillion valuation in its Nasdaq IPO.

Unidentified speaker (speaker_1) ·25:05

How to verify: Check the S-1 filing submitted by Anthropic to the SEC EDGAR database for the IPO prospectus details. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Up to 20% of Anthropic shares are set aside for retail.

Unidentified speaker (speaker_1) ·25:05

How to verify: Check the S-1 filing submitted by Anthropic to the SEC EDGAR database for the IPO prospectus details. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

The FAA went down for the first time since 2023.

Unidentified speaker (UNKNOWN) ·31:03

How to verify: Search news archives from November 2026 for reports on FAA system outages and their dates. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

The Optimizely injection affected Ledger, Zoom, HSBC, and the BBC.

Unidentified speaker (speaker_3) ·35:20

How to verify: Check cybersecurity reports from November 2026 detailing the Optimizely breach and affected clients. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

The Optimizely injection netted about $7,000.

Unidentified speaker (UNKNOWN) ·36:11

How to verify: Check cybersecurity reports or law enforcement statements regarding the financial impact of the Optimizely breach. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

DPRK uses AI to adjust social engineering scripts.

Unidentified speaker (UNKNOWN) ·42:55

How to verify: Search for cybersecurity reports or UN sanctions committee reports from 2026 detailing DPRK's cyber activities and AI usage. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Elon Musk told Speaker_1 in 2018 he would sell a Tesla Roadster that does zero to sixty in under two seconds.

Unidentified speaker (speaker_3) ·41:42

How to verify: Check Tesla product announcements or press releases from 2026-2026 regarding the Tesla Roadster and its performance specs. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Speaker_3 bought 'ant' in 2017.

Unidentified speaker (speaker_3) ·28:39

How to verify: This is a personal claim by the speaker; verify via public disclosures if the speaker is a regulated entity or via their own public statements. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

The 'ANT' token on Hyperliquid is a 'bucket shop' and does not tie to the stock.

Unidentified speaker (UNKNOWN) ·28:58

How to verify: Check Hyperliquid's documentation or smart contract analysis for the ANT token to verify its underlying asset and governance. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Speaker_0 is paying API rates for Mythos access and using it to audit smart contracts.

Unidentified speaker (UNKNOWN) ·48:47

How to verify: Check Mythos documentation or pricing pages for API access models; verify via speaker's public GitHub or code repositories if they disclose usage. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Frontier-level AI was required to find edge cases in the Balancer contract that human auditors missed.

Unidentified speaker (speaker_0) ·48:36

How to verify: Review Balancer audit reports and post-mortems for the specific vulnerability; check if AI tools were credited in subsequent security analyses. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Speaker_1 suspects Deep Sea is running AI agents attacking their accounts.

Unidentified speaker (speaker_1) ·50:59

How to verify: Check on-chain traces for the attacking addresses; search for public disclosures or blog posts from Speaker_1 or Deep Sea regarding these incidents. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Speaker_3 advises crypto teams to prioritize monitoring and detection over traditional audits.

Unidentified speaker (speaker_3) ·52:33

How to verify: Search for Speaker_3's social media posts, blog articles, or podcast episodes from the last six months containing this specific advice. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

The Balancer contract had edge cases that required AI to find.

Unidentified speaker (speaker_0) ·48:36

How to verify: Review Balancer's official security audits and bug bounty reports to identify the specific vulnerability and the method used to discover it. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify Anthropic's S-1 filing for exact share count, price range, and retail allocation details.
  • Compare Anthropic's Opus 5.5 performance metrics against OpenAI's Astra 6.2 and Meta's Muse using independent benchmarks like LMSYS Chatbot Arena.
  • Review Google's Q3/Q4 earnings reports for AI subscription revenue growth and ad model shifts.
  • Check Hyperliquid for 'ANT' token contract address and liquidity depth to confirm it is not tied to Anthropic's equity.
  • Analyze the Optimizely plugin source code and recent CVEs to understand the scope of the injection affecting Ledger, HSBC, and BBC.
  • Evaluate DeFi protocols for the presence of immutable circuit breakers or pause functions versus traditional immutable contracts.
  • Research AI audit firms like Mythos for pricing models and accuracy rates compared to traditional human auditors.
  • Assess the effectiveness of continuous red teaming tools in detecting anomalies in smart contract logs.
Full transcriptexpand

0:00Unidentified speaker (speaker_0): AI is the savior here in a lot of ways. Like this

0:02Unidentified speaker (UNKNOWN): is

0:02Unidentified speaker (speaker_0): human slot. The problems that we're running into are code that is written by humans getting hacked by AI because AI is getting good enough to do it now. And we're gonna get to a transition period where like people start actually believing that like the AI can write better code than them.

0:22Unidentified speaker (UNKNOWN): Hey

0:23Unidentified speaker (speaker_1): everyone, I'm K Mark, and welcome to Uneasy Uneasymoney Money because what happens? on chain never stays on chain. Uh, before we start, here's a word from the sponsors that make this show possible.

0:33Unidentified speaker (speaker_2): This episode is brought to you by Nexo, the premier digital assets wealth platform. Earn daily interest on your crypto or borrow funds, all through one platform trusted by more than 10 million clients. Find out more at Nexo.com. Geovestrictions to eligibility and terms apply. Not listening to bits and Vips yet? It's our weekly crypto and macro market show with Ram Alawalia, Austin Campbell, and Chris Perkins. Search Bits and Vips and subscribe.

1:01Unidentified speaker (UNKNOWN): All

1:02Unidentified speaker (speaker_1): right. Hey everyone. I'm here with my co-host, Taylor Monahan, security expert, and Austin Griffith, builder enablement at the Ethereum Foundation. Welcome, guys.

1:11Unidentified speaker (UNKNOWN): What

1:11Unidentified speaker (speaker_1): up, what up?

1:12Unidentified speaker (UNKNOWN): Back

1:12Unidentified speaker (speaker_0): having me.

1:13Unidentified speaker (speaker_1): Welcome back, Tay.

1:14Unidentified speaker (UNKNOWN): Yeah.

1:15Unidentified speaker (speaker_3): I am here. Let's freaking do it.

1:17Unidentified speaker (speaker_1): Let's do it. Let's do it. Um, okay, so our first segment uh this week is, and forgive me for not understanding time zones. Um, my co-hosts are well aware of my time zone-related problems. But um, anthropic uh is pricing its IPO tonight, and trading starts Thursday on NASDAQ. That's Thursday. I have to assume New York time, not my time. I I assume whoever prepared this document did not adjust it to my time zone. But I my clankers, I I'm I tell them like you have to adjust time zones, man. And so now I don't quite trust if something says Thursday, I'm like, wait, who's Thursday? It's actually worse than before.

1:57Unidentified speaker (UNKNOWN): was, Worse.

1:58Unidentified speaker (speaker_3): Yeah.

1:59Unidentified speaker (speaker_1): Now now I'm like, oh, who wrote this? If a human wrote, this, it's definitely Thursday in New York. If an agent wrote it, it could be anytime.

2:08Unidentified speaker (speaker_3): It could

2:09Unidentified speaker (UNKNOWN): be in the future.

2:09Unidentified speaker (speaker_0): You'll you'll know which way it's

2:10Unidentified speaker (UNKNOWN): a good thing. It could be in the future. Yeah, exactly. That's

2:12Unidentified speaker (speaker_1): why I'm like, I don't want to look like an idiot and be like, hey, it's happening on Thursday, and it's actually on Wednesday or something like that. Um, in any case, ANT Aragon Argon Network Token.

2:25Unidentified speaker (speaker_3): Throwback. Bring the

2:26Unidentified speaker (speaker_0): dead. DAOs back. Yeah.

2:28Unidentified speaker (UNKNOWN): I

2:28Unidentified speaker (speaker_3): still have some of those.

2:29Unidentified speaker (speaker_1): Uh me too.

2:30Unidentified speaker (UNKNOWN): Doubt it.

2:32Unidentified speaker (speaker_1): Yeah. Uh yeah. Yeah. You know your OG if you still have a random bag. They distributed the DAO at that at this point, right? Like they they distributed

2:41Unidentified speaker (UNKNOWN): all

2:41Unidentified speaker (speaker_1): eight. Yeah, yeah. So like you can

2:43Unidentified speaker (speaker_3): Nobody told me I'm holding my ant. aunt.

2:46Unidentified speaker (speaker_1): Yeah. Well, you're unfortunately you're a sucker. I'm a mad. Um, so uh so it's been a crazy week. Uh NVIDIA lost its CEO. Maybe he comes back. I don't know. Maybe there's like a comeback arc. Maybe he goes to jail. We don't know yet. Um, but um it uh it also beat estimates, um, which like the quarter is over, so I'm not quite sure.

3:13Unidentified speaker (UNKNOWN): Like

3:14Unidentified speaker (speaker_1): it's a weird. thing, like, hey, we fired our CEO, and also we're we keep winning. Um, but in any case, uh the deal is um the anthropic deal is they're selling eight hundred and twenty-five million

3:31Unidentified speaker (UNKNOWN): uh

3:31Unidentified speaker (speaker_1): class A shares, uh between 70 to 78 dollars.

3:35Unidentified speaker (UNKNOWN): Um they

3:37Unidentified speaker (speaker_1): haven't worked out. unit bias yet apparently over on the nasdaq um they didn't consider 0.00007 of a cent but okay fair enough um and it's uh a 64 billion dollar raise so they're raising 64 billion dollars in cash um which is uh amazing i'm very excited for my continuing claude max subsidies claude ultra Sorry, I should say Claude Ultra subsidies. It's even better. Um, it feels better when your ultras subsidized than your max, I think.

4:10Unidentified speaker (UNKNOWN): It's

4:10Unidentified speaker (speaker_0): just not amazing Max.

4:12Unidentified speaker (speaker_1): And the the raise is at 1.4 trillion. Um, so

4:18Unidentified speaker (UNKNOWN): up

4:18Unidentified speaker (speaker_1): to 20% is going to retail as well. Um, and you know, we've talked about this before, right? Uh, this idea of um you can't just get rich without taking risks.

4:29Unidentified speaker (UNKNOWN): Um,

4:30Unidentified speaker (speaker_1): and so there's a big question of like, you know,

4:34Unidentified speaker (UNKNOWN): this

4:34Unidentified speaker (speaker_1): is a risky IPO, right? Like these guys could like destroy the world. The risks in this IPO are like so much better than every other IPO that we've had in maybe ever. Like there was a vague sense um that Google could. even though they were saying, like, don't be evil and all of that stuff, right? Like, there was a vague sense of like Google could destroy the world. Um, you know, they get too smart, they hand it over the whole thing to computers or whatever. Like, they seem like the type of guys that could just like accidentally destroy the world. Um, this feels like the first time where it's like, no, no, like not only could they, but they are like kind of trying to on some level, like they're really trying. hard. Like if you were saying, like, let's roll the dice on will like alien intelligence destroy the world, these guys are like, fuck yeah, give us the dice. Let's go. How many rolls do I get, man? How many rolls do I get? Uh so so uh yeah, I mean, it just feels it feels like this is a very risky IPO, existentially, financially. like it's got all the risks. Um, but you know, you don't make money from fucking parking your money in like treasuries of four percent, man.

5:49Unidentified speaker (speaker_0): The thing that you talked about is the thing that I'm most most concerned about. Not getting paper paper clips, not the end of humanity, but my subscriptions, man. Like,

5:56Unidentified speaker (speaker_1): yeah,

5:57Unidentified speaker (speaker_0): is this gonna change the subscription system? Are subscriptions gonna just be like the starting point and then you pay? API credits, like the change to the structure of like basically the most valuable asset on earth right now, like this intelligence to like it's it's worth more than my car, it's worth more than my house. Like this intelligence is everything to me right now. And yeah, if it flips from subscription to API, I'm gonna have to really like change the way I work, change how I'm using it. So for me, it's like

6:26Unidentified speaker (UNKNOWN): I

6:26Unidentified speaker (speaker_0): don't know what's gonna happen to humanity. I just want to keep my subscription rolling.

6:30Unidentified speaker (speaker_1): Maybe uh maybe they're gonna run some like shareholder incentives, right? Where you

6:35Unidentified speaker (UNKNOWN): get

6:35Unidentified speaker (speaker_1): more credits

6:35Unidentified speaker (UNKNOWN): if you I want to just

6:37Unidentified speaker (speaker_1): yeah, some staking, stake dance. Stake dance. Um, so uh so I agree with you. Like the economics are very interesting here, but this is why it is also um

6:49Unidentified speaker (UNKNOWN): such

6:50Unidentified speaker (speaker_1): an interesting. IPO, right? Because like it is not clear that these guys are going to win. I would I would argue even with, and we I think we'll talk a little bit later about um uh the Astra 6.2 launch, right? But you know, since since Opus 5.5, um, it's actually interesting. It's the first time for me, I don't know about you, uh um, you guys, but it's the first time for me where the soda model, the state-of-the-art model from a lab that is winning, which previously was Opus 5.5 high, was like my, you know, go-to guy. He's the only guy I want to talk to. I don't want if I accidentally find myself talking to one of the other guys, I'm like, get this fucking guy out of here. I want to talk to the main guy. And then Fable 5.2 came out. and I found myself using Fable, but I could actually handle going back to Opus.

7:44Unidentified speaker (UNKNOWN): It's the

7:44Unidentified speaker (speaker_1): first time I've ever, it's the first time we're like the jump was like, you know, this guy's smarter in some other ways, but like most of the time I'm kind of fine. So that begs the question of

7:56Unidentified speaker (UNKNOWN): if

7:56Unidentified speaker (speaker_1): Opus 5.5 is like above the threshold, you know, every other time the smartest guy comes to you, like, oh my god, I cannot believe. how dumb that guy. I thought that he was really smart, but actually he was so dumb. And now I never want to talk to him. My best friend, I never want to talk to him again.

8:13Unidentified speaker (UNKNOWN): This

8:14Unidentified speaker (speaker_1): is the first time where like I haven't felt that. Like two new models, two new state-of-the-art models will come out, and I'm still happy to talk to 5.5. Point five.

8:22Unidentified speaker (speaker_0): Plus one. Totally, same thing. I've I've got a few subscriptions, I've got them way along. wailing away. And something happened where I just think burned through like all of my fable credits. I was doing some audits and some other stuff. And it was it was no big deal. It was like, oh, like whatever. Like switch to opus five point five. He can do it. Like let's just let's go. And it was

8:41Unidentified speaker (speaker_1): previously you're like, I'm guess I'm not working today.

8:44Unidentified speaker (speaker_0): Yeah, yes, exactly.

8:45Unidentified speaker (speaker_1): I would rather not get anything done than have to talk to Opus 4.8. four point eight. You make me talk to that guy. I'm gonna blow my brains out. So yeah, it's it's interesting. So so you know, Anthropic is about to IPO. Um, OpenAI was like, How about how about these apples? And no like no one's really reacted. Like, and it's like uh Astro 6.2 six point two is good. Like, don't get me wrong, it's good. Um it's okay to talk to, not amazing. Um, but it is not a threat to the IPO, right? So these guys feel like they are the only ones who have been able to maintain a lead over like multiple models. Like it's been leapfrog mode for a while, but this is uh this is like great timing for them. Opus 5.5 has been dominating for like six weeks. Fable 5.2 comes out. It's also dominating, but it's also fine to stay with Opus 5.5. Like it's a very good setup into the IPO. Um, they feel like they've got some riz that like the other guys are struggling with. Um have you guys used carbon at all?

9:58It's very good. Also, yeah, like I would say it's like Astra. class, like very smart, um, a bit like more autistic than uh than opus, like says like some weird shit every once in a while, but like good.

10:15Unidentified speaker (UNKNOWN): But

10:15Unidentified speaker (speaker_1): like a really good, a strong model, like very strong. How's the

10:20Unidentified speaker (speaker_0): pricing compared?

10:21Unidentified speaker (speaker_1): You know, it's interesting because it's so hard to like

10:24Unidentified speaker (UNKNOWN): API

10:25Unidentified speaker (speaker_0): versus subscription.

10:26Unidentified speaker (speaker_1): it's so it's so hard. Um, like I've got a Google Ultra subscription, um, which I think is $199 uh a month or something. So, you know, it's it's like a little US, so like it's the same price as the the ultra uh subscription. Um it I kind of need some

10:48Unidentified speaker (UNKNOWN): uh

10:48Unidentified speaker (speaker_1): eval to like test. tokens per you know completed task or whatever. I used to have one. I used to have like a uh little uh like 30-minute

10:59Unidentified speaker (UNKNOWN): task.

10:59Unidentified speaker (speaker_1): Yeah, like a 30-minute task that I would hand the agents with the exact prompt, the exact structure, and kind of see how fast did they do it, how efficient were they. Um and then I was like, fuck that. I like I'm only talking to Opus, so it doesn't even matter. Why am I and I stopped maintaining it? I'm like, I'll never talk to anyone else. Um, so um, so yeah, so I guess like to back to the IPO, back to this setup here. Um, Anthropic feels very dominant.

11:25Unidentified speaker (UNKNOWN): They're

11:25Unidentified speaker (speaker_0): the front frog. There's no more

11:27Unidentified speaker (speaker_1): frog.

11:27Unidentified speaker (UNKNOWN): Yeah.

11:28Unidentified speaker (speaker_1): Yeah. Yeah. Um the you know, Muse um Muse Wonder Wonderful Full 1.0, which like in man, like you had a chance, meta to like rebuild the naming conventions here, and you just fucking fumbled hard. Like the naming. So, first of all, Astra 6. We just skip 6.1, go straight to 6.2.

11:51Unidentified speaker (UNKNOWN): Where's

11:52Unidentified speaker (speaker_0): 6.1? Where is it?

11:53Unidentified speaker (speaker_1): It doesn't matter, man. Don't worry about it. It's just 6.2. It doesn't nothing means anything, right? They're just random strings of letters and numbers. Um, you should be used to that. Zero X. yeah,

12:04Unidentified speaker (UNKNOWN): yeah, yeah,

12:04Unidentified speaker (speaker_0): you

12:05Unidentified speaker (UNKNOWN): know,

12:05Unidentified speaker (speaker_1): six point something. So, so um,

12:08Unidentified speaker (UNKNOWN): so

12:09Unidentified speaker (speaker_1): yeah, so we have this uh this setup where Muse, like the top meta model, um, is I would say very powerful, like Opus 4 class, right? And again, it's it's so hard because like they are. quite spiky. So if you give Muse Wonderful Max, like just the name is so bad, man. Um if you give Wonderful 1.0 Max a very hard task, it does a good job. It does a really good job. Like it's architecturally sound, it has good judgment. management. Um, it's a bit annoying to talk to. Um, so you know, meta is has definitely definitely like made a uh I would I would argue like a massive step change similar to what we kind of saw in like November last year, right? Google has also made a massive step change. Like it's gone from like I don't even have a Google subscription because this shit is so fucking bad to no, know I now have two Google subscriptions because I ran out of carbon and I was like, uh, I kind of want this guy to do some of this stuff over here. Like uh browser

13:22use, really solid. Like um, it knows how to use Chrome. It's definitely been trained on some Chrome traces. I promise you that. So um so so anyway, you've got you've got the four labs, um, and two of them are making like really big inroads. They're like two of the largest companies in the world that's fucking infinite money in Google and Meta. And Anthropic is still ahead of them, but like it's a hugely risky setup. And they're gonna go and raise $64 billion at 1.4 trillion. And you know, for you to like make bank on this, like you need them to get to like five trillion. You need to a clean 4x here in my mind. Um, which they could they could just as easily

14:07Unidentified speaker (speaker_3): could.

14:08Unidentified speaker (speaker_1): They could,

14:10Unidentified speaker (speaker_3): which is wild.

14:12Unidentified speaker (UNKNOWN): Like

14:13Unidentified speaker (speaker_3): that, I mean, it's it's like it's legitimately possible. It's not like this is not out of the realm of possibility. But I think there that there are a certain amount of stars that do have to align. And you you know my concern since day one with this is mostly that they'll shoot themselves in the foot or the face. Like

14:31Unidentified speaker (speaker_1): government like governance,

14:33Unidentified speaker (UNKNOWN): governance

14:34Unidentified speaker (speaker_1): nonsense and EA craziness.

14:37Unidentified speaker (speaker_3): 100%. We're the weird org structure, the weird moral grandstanding, the you know yeah, those are the things that I'm I'm sort of m more most concerned about. Um but if they

14:49Unidentified speaker (UNKNOWN): manage to at least they don't have

14:51Unidentified speaker (speaker_1): our money and they can't they kind of stolen our money and we didn't know about it this whole time. So that's bullish.

14:58Unidentified speaker (UNKNOWN): So

14:58Unidentified speaker (speaker_0): they're the front frog but they

15:00Unidentified speaker (UNKNOWN): fuck

15:01Unidentified speaker (speaker_0): themselves up.

15:02Unidentified speaker (UNKNOWN): Yeah I'm just

15:03Unidentified speaker (speaker_3): I'm just praying that they don't fuck themselves. Like that's and that's a wild it's kind of a wild place to be in, right? Because most of the times like the risk oh Well, let me say this When you read the risk disclosures that they've disclosed, right? It is there are so many things that they're hyper aware of and dramatic about, and right.

15:24Unidentified speaker (UNKNOWN): And

15:25Unidentified speaker (speaker_3): yeah, I'm like completely unfazed by like they're like, we might destroy the world. I'm like, yeah, you might. Okay.

15:30Unidentified speaker (speaker_0): Bullet point three, destroy the world.

15:33Unidentified speaker (UNKNOWN): You

15:33Unidentified speaker (speaker_3): might. I'm okay. Accept I accept that risk, right? The risk that I'm still on the fence about accepting is their own internal structure, weirdness. Again, like just the culture, I think, is not necessarily aligned to a really immediate organization. But

15:53Unidentified speaker (speaker_1): like

15:54Unidentified speaker (UNKNOWN): they keep

15:54Unidentified speaker (speaker_1): doing good stuff. Like how

15:55Unidentified speaker (UNKNOWN): can you win if you're

15:57Unidentified speaker (speaker_3): yeah?

15:57Unidentified speaker (speaker_1): Yeah. So so

15:59Unidentified speaker (UNKNOWN): hey,

15:59Unidentified speaker (speaker_1): walk us through your like,

16:02Unidentified speaker (UNKNOWN): you

16:02Unidentified speaker (speaker_1): know. your governance lens view of the four soda labs, right? Like Meta, Google, OpenAI, Anthrop.

16:11Unidentified speaker (UNKNOWN): Yeah.

16:12Unidentified speaker (speaker_1): Like how like give give us the setup on those four.

16:16Unidentified speaker (UNKNOWN): I

16:16Unidentified speaker (speaker_3): mean, I think if you look at, and I'll throw Apple in this boat too, even though they

16:23Unidentified speaker (UNKNOWN): I

16:23Unidentified speaker (speaker_0): don't know, seriously, great.

16:25Unidentified speaker (speaker_1): Think about

16:25Unidentified speaker (UNKNOWN): that.

16:25Unidentified speaker (speaker_1): Tim Apple is

16:26Unidentified speaker (UNKNOWN): a

16:26Unidentified speaker (speaker_1): asleep at the wheel, man. Like I'm

16:28Unidentified speaker (UNKNOWN): not sure. This is what I mean

16:29Unidentified speaker (speaker_3): though. I think the biggest risk for let me just okay. So you'll see Okay, so Google Meta, it'll throw Apple in his boat, right? I think the biggest risk is like they're so entrenched. They've they've built their thing and then they're trying to do AI too. I think that obviously some are doing better than others. Um notably Google and Meta are doing better than Apple in this regard. But

16:55Unidentified speaker (UNKNOWN): it's hard for

16:57Unidentified speaker (speaker_1): the gap was big, right? And they've closed the gap quickly. So

17:00Unidentified speaker (UNKNOWN): that's

17:00Unidentified speaker (speaker_1): that's fair. That's bullish

17:02Unidentified speaker (UNKNOWN): for

17:02Unidentified speaker (speaker_1): Apple.

17:04Unidentified speaker (speaker_3): And that's and that's the thing is like, you know, it's if you've they can overcome the existing entrenched culture and organization and priority. priorities and figure out a way to really ship like amazing AI, then um then they're sort they're in a better position to do so theoretically. It's just a matter of like actually shipping something that's valuable for people. Right.

17:31Unidentified speaker (speaker_1): So so maybe again to like ask the question uh in a different way, right? Like you need humans still, right?

17:42Unidentified speaker (UNKNOWN): Like you

17:42Unidentified speaker (speaker_1): need smart researchers doing research, right? And you need a fuckload of money and access to compute.

17:48Unidentified speaker (UNKNOWN): Yeah.

17:49Unidentified speaker (speaker_1): Like those are the primary inputs, right, into this process at the moment. Um, like Apple obviously has

17:57Unidentified speaker (UNKNOWN): uh

17:58Unidentified speaker (speaker_1): very smart engineers, right? Although, you know, in AI, like I don't know what the fuck they those guys have been doing. Um, but like,

18:06Unidentified speaker (UNKNOWN): do

18:06Unidentified speaker (speaker_1): you do you just see it as like this hyperscaler race of you can sort of acquire the ladder? You can acquire like smart people. Like Meta did. They

18:14Unidentified speaker (UNKNOWN): were

18:15Unidentified speaker (speaker_1): like, We'll pay you a hundred billion trillion d dollars to come and

18:19Unidentified speaker (speaker_3): do

18:19Unidentified speaker (speaker_1): our thing. And they got enough people, clearly, that they were able to close whatever gap. Like

18:25Unidentified speaker (speaker_3): I think that it is a huge battle to actually like train and ship the model. And the model has to be competitive, right? Like I think that that's like your sort of bare minimum. But I do think that there is an additional layer that um that is probably the most human thing, right? Which is like the decision making and the positioning of the product that you're shipping and how people are using it and interacting with it. And then also how your organization. is making decisions and interacting with the world, right? So interacting with researchers, interacting with uh government, literally the US government, um, you know, all of these sorts of things. And if they cannot get uh get in a good position, I think that as we've seen with Apple, right, if they

19:13Unidentified speaker (UNKNOWN): just

19:13Unidentified speaker (speaker_3): cannot put to two and two, like you

19:15Unidentified speaker (UNKNOWN): can

19:15Unidentified speaker (speaker_3): hire again, you can hire all the best people in the world, but you the stars do have to align, the things have to click in place. Um and I think it's very interesting to look at like Google Meta on one side and then anthropology

19:27Unidentified speaker (UNKNOWN): on the other. Like how do you

19:29Unidentified speaker (speaker_1): lay out the spectrum here of those four? Like where on the spectrum

19:33Unidentified speaker (UNKNOWN): do

19:34Unidentified speaker (speaker_1): you in terms of like maybe governance coherence or or whatever?

19:40Unidentified speaker (speaker_3): I think that Google and Meta's like biggest challenge is how do you push how do you take an existing company, an existing org structure, and these are huge organizations, right? And how do you prioritize AI and ship AI to your users in a way that creates value for your users, but then you know obviously is valuable for you. And Anything I think like Google, the the sort of like stereotypical example is Google, like they don't want to cannibalize ad revenue and just like how do you

20:09Unidentified speaker (UNKNOWN): seem like

20:10Unidentified speaker (speaker_1): they've maybe capitulated on that at this point. Right. Said, like, fuck, if we don't destroy the world,

20:15Unidentified speaker (UNKNOWN): we don't do this, get

20:16Unidentified speaker (speaker_1): there before us.

20:17Unidentified speaker (speaker_3): Exactly. And they were able to do that, right? They were able, like they maybe they were a little bit slow to the punch or whatever, but they got there. I mean,

20:24Unidentified speaker (speaker_1): they have

20:25Unidentified speaker (UNKNOWN): had deep

20:25Unidentified speaker (speaker_1): mind, it's fucking wild. They were the guys that wrote the paper. They had deep mind. Like remember

20:31Unidentified speaker (UNKNOWN): those cool

20:32Unidentified speaker (speaker_0): visualizations? Those were amazing.

20:34Unidentified speaker (speaker_1): Like Demis was like off like folding proteins or whatever. And like clearly they had to be like, bro, like fold some code, man. Like, let's fucking let's start folding some stuff that we can sell.

20:49Unidentified speaker (UNKNOWN): And

20:49Unidentified speaker (speaker_0): ad revenue already has to be tanking. right? Aren't people talking to their agents? Like, don't you add like no one's going to be Googling anymore? Like it doesn't

20:56Unidentified speaker (UNKNOWN): even make

20:57Unidentified speaker (speaker_0): sense to Google.

20:59Unidentified speaker (speaker_3): Well, and my I'll say that my biggest interactions with Google's uh AI is now Google search results. And I finally like connected it to my primary Google account. I finally bit the bullet and let it do it. And it's it's quite good. Cause I can just type it in the search. bar. I just type like a random question. And then the sort of AI results are actually pretty they're

21:22Unidentified speaker (UNKNOWN): they were annoying.

21:23Unidentified speaker (speaker_3): They

21:24Unidentified speaker (UNKNOWN): were

21:24Unidentified speaker (speaker_3): annoying at first. I was like,

21:25Unidentified speaker (UNKNOWN): get out of

21:26Unidentified speaker (speaker_3): here.

21:27Unidentified speaker (UNKNOWN): But they've

21:27Unidentified speaker (speaker_3): gotten so much better. And it's almost like that top little AI dialogue. And then it the way that they've like positioned the links within that.

21:36Unidentified speaker (speaker_1): Dude, let

21:37Unidentified speaker (UNKNOWN): me tell you it's turning

21:38Unidentified speaker (speaker_3): into search. Like that's

21:39Unidentified speaker (UNKNOWN): what search is going to

21:40Unidentified speaker (speaker_3): be.

21:41Unidentified speaker (speaker_1): If you connected your account and had carbon as like the

21:44Unidentified speaker (UNKNOWN): guide

21:45Unidentified speaker (speaker_1): you could then ask questions to, your mind would be blown. I'm not kidding.

21:48Unidentified speaker (UNKNOWN): Oh, so I

21:49Unidentified speaker (speaker_3): so I should just I should just go all in. I should just dive in at first. Is that what

21:52Unidentified speaker (UNKNOWN): I'm saying? So

21:54Unidentified speaker (speaker_1): like like if you like the the those AI results that they're presenting to you, um they're clearly throwing more. compute at that and like throwing better models at it. Like that's the kind of result that you get from from carbon if you ask a weird, especially if you ask carbon a question that involves like breadth.

22:15Unidentified speaker (UNKNOWN): It

22:15Unidentified speaker (speaker_1): it comes back very like it's it's less spiky, it feels like, in terms of its capabilities. Um

22:21Unidentified speaker (speaker_3): Interesting.

22:22Unidentified speaker (speaker_1): But yeah, yeah. Uh I think, you know, the other day I a weird thing where like I had a this pump fucking clue, right? And I was like, I need to get a new pump. And and I'm used to just going to Claude if I want to get something. Usually it's information. Like I don't buy like weird shit that I don't know where to buy it very often. And so I was like, hey, I need to get this pump. And honestly, it was a fucking terrible experience. It was like, oh, buy buying it on eBay. Or like maybe like AliExpress or whatever. And I was like, what do you mean, maybe AliExpress? Like, can't you search for? Like, what are you doing, man? Like, it was it was really bad to the point where I was like, all right, I'm gonna put my phone down and go and actually go to my harness and be

23:04Unidentified speaker (UNKNOWN): like,

23:04Unidentified speaker (speaker_1): hey, bro, I need you to help me find this thing. Um so like Claude definitely has some gaps that Google,

23:11Unidentified speaker (UNKNOWN): you

23:12Unidentified speaker (speaker_1): know, I don't have the the I actually have the anti-gravity app. I didn't even think to use it, which is pretty bearish. Um but um

23:19Unidentified speaker (UNKNOWN): the

23:20Unidentified speaker (speaker_1): Google

23:20Unidentified speaker (speaker_3): Google the Google interaction for me personally, the Google interaction is for right now it's just a hundred percent like distribution based like meaning that like I type it in the Chrome search

23:31Unidentified speaker (UNKNOWN): bar. It's like

23:32Unidentified speaker (speaker_3): I don't even consider it

23:35Unidentified speaker (UNKNOWN): go switching to the clock AI.

23:37Unidentified speaker (speaker_1): Yeah. It's

23:38Unidentified speaker (UNKNOWN): just like

23:38Unidentified speaker (speaker_3): I know need a thing, I want an answer in the same way that I'm like, what's the weather today? Whatever, I need this thing. How do you change a faucet head? Whatever, you know, like I used to go to Google, I still go to Google. And I'm

23:51Unidentified speaker (UNKNOWN): not sure. Which I should change my thing.

23:52Unidentified speaker (speaker_0): Like you're not gonna go to Google and scroll through the results

23:56Unidentified speaker (UNKNOWN): like you

23:57Unidentified speaker (speaker_0): used to, but maybe you go there because it's the place that you're used to asking questions and

24:01Unidentified speaker (UNKNOWN): they

24:01Unidentified speaker (speaker_0): get

24:02Unidentified speaker (UNKNOWN): better at

24:02Unidentified speaker (speaker_0): answering. with AF.

24:03Unidentified speaker (UNKNOWN): Yeah,

24:04Unidentified speaker (speaker_1): the Google box was just like going to Claude and you type type, you know. The problem is the scale of that. Like Google scale, even for Google, they can't afford to be giving away free chat bot sessions. Like even Google can't afford that. Um so you need to have some kind of subscription, which then means you've got this like dual mode. Like the nice thing about Google was it was free for everyone. You could just type in a thing and it just worked and the ads paid for it. Um, I'm sure they'll they're pretty good at ads. I'm sure they'll

24:33Unidentified speaker (speaker_3): I think they'll figure I think they'll figure it out. But

24:35Unidentified speaker (UNKNOWN): these answers are

24:35Unidentified speaker (speaker_0): cash too, right? It's sort of like if one person asks one good question, it goes to AI one time to get a good answer,

24:42Unidentified speaker (UNKNOWN): and

24:42Unidentified speaker (speaker_0): then it's kind of there, right? And then if anybody else asks that same question, it's not doing

24:46Unidentified speaker (UNKNOWN): you can cache it and refresh, yeah.

24:49Unidentified speaker (speaker_3): Especially for questions like what is the shower head or where you know like how do you attach

24:53Unidentified speaker (UNKNOWN): the

24:53Unidentified speaker (speaker_3): shower head to the wall or like you know these things that like uh you know when we're dealing with the scale of Google you

24:59Unidentified speaker (UNKNOWN): guys are both playing in the plumber

25:01Unidentified speaker (speaker_0): beam I

25:01Unidentified speaker (speaker_1): don't know what

25:02Unidentified speaker (UNKNOWN): we're doing the

25:02Unidentified speaker (speaker_1): plumber I'm trying to I'm my HVAC uh career

25:08Unidentified speaker (UNKNOWN): oh

25:08Unidentified speaker (speaker_0): air pump obviously no

25:11Unidentified speaker (speaker_3): no I'm not

25:11Unidentified speaker (UNKNOWN): on the shower. mine actually wasn't water pump it

25:13Unidentified speaker (speaker_1): was a cooling system one water pump. So

25:16Unidentified speaker (speaker_3): my shower head is. Yeah, I got a new shower head. I had to figure out how to put it on the can't be bothered to have a plumber come over. The AI is right here, guys. What are you doing?

25:26Unidentified speaker (UNKNOWN): And

25:26Unidentified speaker (speaker_0): wait till we add your battery in. And you can just

25:28Unidentified speaker (UNKNOWN): send the dude down

25:29Unidentified speaker (speaker_0): to

25:29Unidentified speaker (UNKNOWN): the

25:29Unidentified speaker (speaker_1): this was the other thing though. The funny thing was I was trying to buy this stupid pump, right? And Claude was better. This is quite hilarious, actually, because I was

25:37Unidentified speaker (UNKNOWN): like,

25:38Unidentified speaker (speaker_1): hey, I just need to buy this pump. I took a photo and it's like, oh yeah, that's the knockoff Chinese version. I was like, excuse me, what are we what's going on here? It was like, oh, well, there's like the like original industrial version. I don't know anything about pumps, but I went down a pump fucking rabbit hole where I was like, Explain to me why. You Explain to me why this is. And it was like, Well, you see, the ceramic disc that the thing uses. for its like, and I was like, Okay, let's tell me more. I'm like, now I'm like so deep in pump lore of like what and I was like, but I still don't understand why this pump costs eight hundred dollars and the knockoff version costs eighty dollars. And it was like, Well, they put acid through the other one, and it's totally fine, like you'll never buy another. It was selling me the pump really well. It was like, you'll never, I'm like, I'm like I'm

26:27Unidentified speaker (UNKNOWN): not moving a lot of acid around acid

26:29Unidentified speaker (speaker_1): through

26:29Unidentified speaker (UNKNOWN): this pump. Yeah

26:30Unidentified speaker (speaker_1): like no I don't need to put acid through the fucking pump. But I was like I kind of want the acid pump. It seems cooler

26:38Unidentified speaker (speaker_3): just in case

26:39Unidentified speaker (UNKNOWN): like

26:40Unidentified speaker (speaker_1): sulfuric acid just like spraying around the whole

26:45Unidentified speaker (speaker_3): you never know when these things come in handy.

26:47Unidentified speaker (speaker_1): Yeah. If I ever switch to acids in the in this pump I don't want to be I don't have to order a whole other pump. That'll be crazy. I might as well just use the good one. So um but yeah, you had to you in order to buy the like proper industrial one, you had to like fucking speak to a human. It's like still like a whole thing. And I was like, oh

27:07Unidentified speaker (speaker_3): yeah, it's

27:07Unidentified speaker (speaker_1): not worth it. I'm

27:09Unidentified speaker (UNKNOWN): that's I'm buying the normal

27:10Unidentified speaker (speaker_3): one. This is why I put my shower head up. Okay. I didn't want to speak say it's to so a human. human

27:14Unidentified speaker (speaker_1): i

27:15Unidentified speaker (UNKNOWN): i don't have him in

27:16Unidentified speaker (speaker_3): my house i

27:17Unidentified speaker (speaker_1): know right um all right so let's let's wrap up this this thing we'll see um you know i've got uh i've got a bunch of um people that have only been talking about this this i po for the last like two weeks basically and i think there's a bunch of people that are pattern matching on the spacex ipo it's gonna pump and then it'll be a dip and there's a buy buying opportunity, etc. etc. Like, have they repriced for the market being bearish?

27:44Unidentified speaker (UNKNOWN): You know,

27:44Unidentified speaker (speaker_1): it's a good buy under a trillion or whatever. Like, I I don't know, man. Um, they're trying to destroy the world. Feels like feels like a must-buy. You at least you know what? How like you know, you remember the quote of like, how can you wake up the the D DJen Jen Spartan quote, right? Like, um, you don't want to wake up every day, uh, you know, to the industry. that you like read every message, whatever. I can't I can't remember the exact quote, but um, and not be exposed to it. Like, what the

28:12Unidentified speaker (UNKNOWN): fuck are

28:12Unidentified speaker (speaker_1): you doing? How can you not have how can you not own tokens if you like wake up every day and do crypto stuff for 14 hours and you don't have exposure to it? I feel like how can you not

28:24Unidentified speaker (UNKNOWN): have

28:24Unidentified speaker (speaker_1): exposure to the guys that might destroy the place that you live? Like at least make some money on the way out, right? Like, you

28:31Unidentified speaker (UNKNOWN): don't

28:31Unidentified speaker (speaker_0): want to be

28:31Unidentified speaker (UNKNOWN): how dumb would you

28:32Unidentified speaker (speaker_1): feel if they destroy the world and you're like, I didn't even make any money from this shit. Like

28:38Unidentified speaker (UNKNOWN): Yeah. This

28:39Unidentified speaker (speaker_3): is why

28:39Unidentified speaker (UNKNOWN): I bought

28:40Unidentified speaker (speaker_3): ant this is why I bought ant in 2017.

28:42Unidentified speaker (UNKNOWN): Right. The

28:43Unidentified speaker (speaker_3): old ant, right? And

28:44Unidentified speaker (speaker_1): yeah,

28:45Unidentified speaker (speaker_0): there's another ant, right? There's an ant trading on hyperliquid right now, right? So there's

28:49Unidentified speaker (UNKNOWN): the

28:49Unidentified speaker (speaker_0): old ant, there's the fake

28:50Unidentified speaker (UNKNOWN): ant.

28:51Unidentified speaker (speaker_0): and

28:52Unidentified speaker (UNKNOWN): what

28:52Unidentified speaker (speaker_0): what is that ant? Like, how does that even work? Does it even tie to the stock at all? Can you buy

28:58Unidentified speaker (UNKNOWN): the stock

28:58Unidentified speaker (speaker_0): now, really? No.

28:59Unidentified speaker (UNKNOWN): No, no, you're just it's

29:01Unidentified speaker (speaker_1): you're just it's a bucket shop. It's

29:03Unidentified speaker (UNKNOWN): like

29:03Unidentified speaker (speaker_0): trust me. Back to the buckets. Okay.

29:05Unidentified speaker (UNKNOWN): Yeah.

29:05Unidentified speaker (speaker_1): Got

29:05Unidentified speaker (speaker_0): it.

29:05Unidentified speaker (speaker_1): Back to buckets. Back to buckets. Um everything's a bucket shop.

29:09Unidentified speaker (UNKNOWN): If

29:10Unidentified speaker (speaker_1): you don't know what it is, it's a bucket shop. Don't worry about it. All right. Let's go to ads. Uh good luck with your IPO buying. Um, we'll be back in a sec.

29:18Unidentified speaker (speaker_2): What if your crypto could do more than just sit in your wallet? Nexo is a digital assets wealth platform where you can generate daily interest on your crypto or borrow funds against your assets. Today, Nexo serves more than 10 million clients. Find out more at Nexo.com.

29:37Unidentified speaker (UNKNOWN): If you

29:38Unidentified speaker (speaker_2): like this show, there's another one from our network you should be listening to. Bits and Bips is our weekly market show. On Mondays, Rom Alawalia, Austin Campbell, and Chris Perkins chat about where crypto and macro collide. It's sharp, it's fast, and it has its own channel. Search Bits plus sign bips and subscribe.

30:00Unidentified speaker (speaker_1): All Alright, right. um, so it was a bad week in everything, basically. Um unless you're raising a bunch of money from the stock market, you've had a rough week. Um we had Microsoft go down. Um I saw so many great jokes about Microsoft going down. Someone was like someone was like, uh I saw a tweet that It was like, It's been a bad week to be a co co-pilot. pilot.

30:24Unidentified speaker (speaker_0): Oh

30:25Unidentified speaker (UNKNOWN): that's

30:25Unidentified speaker (speaker_0): like FAA and everything. right?

30:27Unidentified speaker (UNKNOWN): FAA, yeah, yeah. Yeah.

30:29Unidentified speaker (speaker_1): Um so so FAA went down. They lost the frequencies. This is my favorite explanation for an outage ever. The frequencies stopped working, guys.

30:41Unidentified speaker (speaker_0): That's like old school made up, like physics breaking or something, right?

30:45Unidentified speaker (UNKNOWN): Yeah,

30:46Unidentified speaker (speaker_1): just the there was no we didn't we ran out of frequencies for a minute there. Um so yeah, so so Microsoft had like two different outages. Um the FAA for the first time since 2023 was like down. Um

31:03Unidentified speaker (UNKNOWN): and

31:03Unidentified speaker (speaker_1): then the previous time before that was like 9 11. Nine

31:05Unidentified speaker (UNKNOWN): eleven

31:06Unidentified speaker (speaker_1): rounded everything.

31:06Unidentified speaker (UNKNOWN): Oh, yeah.

31:07Unidentified speaker (speaker_1): Um so then they're like

31:09Unidentified speaker (UNKNOWN): oh

31:10Unidentified speaker (speaker_3): I thought the FAA one was gonna be way, a way different thing than what it was. Like I

31:15Unidentified speaker (UNKNOWN): thought I

31:16Unidentified speaker (speaker_3): was like it was

31:16Unidentified speaker (UNKNOWN): like it

31:17Unidentified speaker (speaker_3): was like 911 throwback. I was like, oh shit,

31:19Unidentified speaker (UNKNOWN): do this. I know

31:20Unidentified speaker (speaker_1): it was yeah, same.

31:21Unidentified speaker (UNKNOWN): I just got done flying.

31:22Unidentified speaker (speaker_0): I had a cold. I'm like, oh man, they're about to announce that I'm gonna die.

31:25Unidentified speaker (UNKNOWN): Man, I

31:26Unidentified speaker (speaker_0): know it. I know it.

31:28Unidentified speaker (speaker_3): And then it was like it was like, oh no, it's just it's just broken. Like everyone.

31:33Unidentified speaker (UNKNOWN): The

31:34Unidentified speaker (speaker_1): radio's broke. Sorry guys. So

31:35Unidentified speaker (UNKNOWN): I

31:35Unidentified speaker (speaker_3): guess that's

31:36Unidentified speaker (UNKNOWN): it. You can't

31:36Unidentified speaker (speaker_0): use floppy disks.

31:37Unidentified speaker (UNKNOWN): I

31:38Unidentified speaker (speaker_1): did hit I did. see the the there was a funny thing where like um apparently when they connected the new shit to the old shit, um and it was like doing like a thousand transactions a second or something like that. The like it was it was like hilariously slow and the system just bogged down and and shut down. So um then we had a bunch of banks and other shit. happen. And then the scariest one, genuinely to me, the scariest one, because this happens so infrequently, um, was Apple.

32:11Unidentified speaker (UNKNOWN): Um the

32:14Unidentified speaker (speaker_1): Apple one is like they usually have their shit together. They don't, they don't, they're pretty like mission critical, careful with stuff. I don't know if we have an explanation yet of what happened with Apple. Um but um

32:29Unidentified speaker (UNKNOWN): yeah

32:29Unidentified speaker (speaker_3): it's been really hard for me to gauge the reality of all these different situations.

32:35Unidentified speaker (UNKNOWN): Yeah

32:35Unidentified speaker (speaker_1): what's your take here Tay like these are like

32:38Unidentified speaker (speaker_3): dude it's

32:39Unidentified speaker (UNKNOWN): downtime

32:40Unidentified speaker (speaker_1): but they're not like hacks

32:42Unidentified speaker (speaker_3): and it's just and then you

32:45Unidentified speaker (UNKNOWN): it's

32:45Unidentified speaker (speaker_3): hard to too because you have it used to be back in the olden days. It used to be you just had to sort of overcome the PR spin. So like Apple would say something and you sort of had to like drill into it, or Microsoft would make a statement drill into it to get to like the, you know, probably the proper truth. And like journalists would help with this, and social media would help with this, and whatever. These days, though, I just feel like I'm ha like you haven't you have this weird layer where like the reality is sort of AI. slop and that's what you have to penetrate through and then you have like a PR statements like over here just like chill in and they have almost no relevance anymore. Like it's super it's just super weird. Um I think I asked in some of my like trusted group chats like four separate times over the last week the optimizely Optimiz optimizely Optimizely?

33:35Unidentified speaker (speaker_1): optimizedly Optimizely. yep yeah

33:37Unidentified speaker (speaker_3): that incident I had to ask I was like, wait, so did they just screw like they screwed themselves or what? Because that was like a okay. So let's let's this one's interesting. Optimizely is like a plugin that so many websites use.

33:50Unidentified speaker (speaker_1): I used to use this on my e-commerce site. Like

33:53Unidentified speaker (UNKNOWN): 100%.

33:54Unidentified speaker (speaker_1): It had just come out when when this is like 2012. Yeah.

33:59Unidentified speaker (speaker_3): And it is it is um for like Google Analytics is right. It's just like a thing you put it on your site and it helps you with A B testing and

34:10Unidentified speaker (UNKNOWN): also like you have analytics. Like a

34:13Unidentified speaker (speaker_1): hot uh I I can't remember what the feature was, but this is one of the coolest features, especially back in 2012. Um, you could see the tracking of

34:22Unidentified speaker (UNKNOWN): like the

34:22Unidentified speaker (speaker_1): heat

34:22Unidentified speaker (UNKNOWN): map cursor like yeah

34:25Unidentified speaker (speaker_1): through the like a live a heat

34:26Unidentified speaker (UNKNOWN): map.

34:27Unidentified speaker (speaker_1): of like where people were on the side where they were looking

34:29Unidentified speaker (UNKNOWN): a hundred percent.

34:30Unidentified speaker (speaker_1): It's so

34:31Unidentified speaker (speaker_3): yeah. So it's not just like I yeah the heat map was okay. So basically you have like your website or your app or whatever. It'll basically collect all this data and then it puts it into a really fancy visualization where you can actually understand where your users are like dropping off, where they're more interested, what's hitting, what's not hitting and that's how you improve your app, right? If you're seeing people sort of primarily using one feature or whatever, you want to like hide.

34:56Unidentified speaker (UNKNOWN): I mean, e-commerce is

34:57Unidentified speaker (speaker_1): like the ultimate like use case for

34:59Unidentified speaker (UNKNOWN): this, right? Because like yeah,

35:01Unidentified speaker (speaker_3): you're optimizing the hell out of your checkout flows, right? You have like six percent make it through the first click, and six percent of those make it through the second, etc. etc. So this is like one of the best tools, it's been around forever, and therefore it just lives in all of these different websites, right? Like SaaS products, crypto products, non-crypto products. Uh what's the list? Ledger, Zoom, HSBC, and the BBC. Like everyone uses this thing, right? And then somehow Now someone uh hacked it, and so they just did a little injection. So now the hackers were. basically like in sort of all these different websites.

35:43Unidentified speaker (speaker_1): You've got fucking JavaScript running in your website, man.

35:45Unidentified speaker (UNKNOWN): Like

35:46Unidentified speaker (speaker_3): right. Um, and so I was like, Oh, that's really bad because it's the breadth of the organizations that are impacted by this are again crypto, non-crypto, banks, non-banks, intact

35:58Unidentified speaker (UNKNOWN): it's

35:58Unidentified speaker (speaker_0): like the little

35:59Unidentified speaker (UNKNOWN): skimmer, like

36:00Unidentified speaker (speaker_0): when you see a little skimmer, like the little piece of hardware

36:02Unidentified speaker (UNKNOWN): someone.

36:03Unidentified speaker (speaker_0): puts on top of a uh credit

36:05Unidentified speaker (UNKNOWN): card

36:05Unidentified speaker (speaker_0): machine, but it's but it's but it's on like

36:08Unidentified speaker (UNKNOWN): 160 million

36:09Unidentified speaker (speaker_0): websites all at once. Yeah, but there

36:11Unidentified speaker (UNKNOWN): wasn't much

36:11Unidentified speaker (speaker_0): money taken, right? I see seven grand in here when I'm looking at that. How did they only get seven grand? What a fumble.

36:16Unidentified speaker (speaker_3): I went I went to four different group chats. I was

36:19Unidentified speaker (UNKNOWN): like,

36:20Unidentified speaker (speaker_3): can someone fact check this, please? Because did they drop the ball or what? How do you inject yourself on this? breadth of stock and take seven thousand dollars total? Like I'm sorry, like a literal kid doing like a basic crypto phishing site makes more than seven thousand dollars. What is happening?

36:40Unidentified speaker (UNKNOWN): But

36:41Unidentified speaker (speaker_1): see Tay, right? Like correct me if I'm wrong. One of the uh nice things, maybe about lowering the barriers to entry of hacking everything in the world, is most of the people who are now hacking things have no skills whatsoever and have no plan.

36:59Unidentified speaker (speaker_3): So that and that might be it reminds me a bit of the Twitter incident um years ago. I don't know if you

37:04Unidentified speaker (UNKNOWN): remember

37:04Unidentified speaker (speaker_3): this one, right? The kids, the kids got full unlimited access to Twitter. They could take over any Twitter account. They took over, in fact, they took over like Coinbase, they took over Obama, they took over the Apple account that had never tweeted before. Oh. And you're like, all right, World War III is happening. Like they had they rule the world, right? They are they are Obama. They can tweet as Obama. What do they do? They do a uh send us one Bitcoin and we'll send you two Bitcoin back campaign.

37:34Unidentified speaker (UNKNOWN): What the hell? You

37:37Unidentified speaker (speaker_3): need uh I'm not I think they maybe like a hundred and sixty thousand dollars. Yeah

37:40Unidentified speaker (speaker_1): it was like

37:41Unidentified speaker (UNKNOWN): a hundred

37:41Unidentified speaker (speaker_1): grand. It was the craziest thing. It was like the the like

37:44Unidentified speaker (UNKNOWN): a

37:45Unidentified speaker (speaker_1): generational fumble. Those guys must I don't know if they're in jail. I don't know if they got caught.

37:49Unidentified speaker (UNKNOWN): They are

37:49Unidentified speaker (speaker_3): in jail. You

37:50Unidentified speaker (UNKNOWN): don't take

37:50Unidentified speaker (speaker_3): over Obama's Twitter and not go to jail.

37:53Unidentified speaker (speaker_0): Yeah, fair.

37:54Unidentified speaker (speaker_1): Fair.

37:55Unidentified speaker (UNKNOWN): Fair. You wouldn't

37:56Unidentified speaker (speaker_3): started World War Three, Or bro. And yeah,

37:58Unidentified speaker (UNKNOWN): but you know what?

37:59Unidentified speaker (speaker_3): Congrats on your crypto giveaways. Yeah. Okay. Good

38:01Unidentified speaker (speaker_1): job.

38:02Unidentified speaker (speaker_3): Anyways. that's what I feel like the optimizedly one is. I heard that they might that they

38:07Unidentified speaker (UNKNOWN): they

38:07Unidentified speaker (speaker_3): sort of had like a variety of little skimmer things. So maybe it just wasn't crypto focused or a crypto focused red actor.

38:14Unidentified speaker (speaker_1): I don't

38:14Unidentified speaker (UNKNOWN): think it's just bizarre.

38:16Unidentified speaker (speaker_3): It's just

38:16Unidentified speaker (speaker_1): optimizedly is is a much more prevalent thing in uh e com than it is in in crypto. Like I I I don't even know, I can't think of a single portfolio company. that I know of that uh if I asked them about it, they would even know about it. Like it's never even come up in a conversation. Um uh so I think maybe we dodged a bullet there. Um yeah,

38:37Unidentified speaker (speaker_3): a hundred percent. I feel like we've dodged a lot of bullets up. Like FAA could have been

38:41Unidentified speaker (UNKNOWN): worse than Microsoft could have been worse. Like, what

38:43Unidentified speaker (speaker_3): is happening?

38:43Unidentified speaker (speaker_1): This is sort of what I'm saying, right? Like, you know, if you give guns to a bunch of kids. they're gonna like shoot some people, but like mostly miss. They're not trained assassins.

38:57Unidentified speaker (UNKNOWN): I

38:57Unidentified speaker (speaker_1): don't know that it's good, but like it'd be worse if the guns were like given to trained assassins. Those guys aren't gonna miss, they're gonna they're gonna fucking kill a bunch of people. So I don't know.

39:12Unidentified speaker (UNKNOWN): We shouldn't give guns

39:12Unidentified speaker (speaker_1): to kids.

39:14Unidentified speaker (UNKNOWN): I My

39:16Unidentified speaker (speaker_0): last disclaimer.

39:18Unidentified speaker (speaker_1): Disclaimer. Don't don't give guns to

39:20Unidentified speaker (UNKNOWN): kids.

39:21Unidentified speaker (speaker_1): Um

39:21Unidentified speaker (UNKNOWN): but so

39:22Unidentified speaker (speaker_1): actually,

39:22Unidentified speaker (UNKNOWN): you know

39:23Unidentified speaker (speaker_1): what's crazy? Actually, I should I should say this. I was in New Zealand skiing and we had a bunch of kids with us and I took them to a shooting range. And actually, like not even joking, they're super accurate. Like they're actually really dangerous.

39:37Unidentified speaker (speaker_3): So

39:37Unidentified speaker (UNKNOWN): actually it's okay to

39:38Unidentified speaker (speaker_3): give them guns.

39:39Unidentified speaker (UNKNOWN): Dude, one

39:39Unidentified speaker (speaker_1): kid was just like blowing these like play pigeons out of the air. Like they've been playing fucking like first person

39:46Unidentified speaker (UNKNOWN): shooter

39:46Unidentified speaker (speaker_1): games. They're

39:47Unidentified speaker (UNKNOWN): like video games, and it's about to say, yeah.

39:50Unidentified speaker (speaker_1): Do not give do not give kids guns. They they're actually maybe worse than assassins. I changed

39:54Unidentified speaker (UNKNOWN): my whole

39:55Unidentified speaker (speaker_0): terrible analogy

39:56Unidentified speaker (UNKNOWN): habitually.

39:57Unidentified speaker (speaker_1): It's a terrible analogy. Children are amazing at shooting guns, and we shouldn't give them to them. Uh but

40:02Unidentified speaker (UNKNOWN): with all these

40:03Unidentified speaker (speaker_0): hacks, this make this reminds me that like we need to rip the band-aid off, or the band-aid is currently being ripped off. But we're in this transition period. Even the FAA hack, there were so many people posting tweets where they were juxtaposed with other tweets of FAA adding AI stuff. But this had nothing to do with AA. AI. AI was the attacker. AI is the savior here in a lot of ways. Like this is human slop. The problems that we're running into are written code that is written by humans getting hacked by AI because AI is getting good enough to do it now. And we're gonna get to a transition period where like people start actually believing that like the AI can write better code than them, but we're still in this period movie. of

40:45Unidentified speaker (UNKNOWN): the state.

40:45Unidentified speaker (speaker_0): We still doesn't

40:46Unidentified speaker (UNKNOWN): believe that. A lot of people a lot of people know. no one

40:51Unidentified speaker (speaker_1): believes that. That's

40:53Unidentified speaker (speaker_3): well, it's weird because I don't know. There are people though that they have this. I think what happened, so for the intelligent, like the intelligent people who should know better but don't know better, I think what happened is that they were sort of like sold this thing, this vision, this very optimistic and dramatic sort. of version of reality. And then when they actually looked, they were like, okay, this is just like, you know, this is not true whatsoever. And they've sort of written off everything that they've heard. We've seen this a lot with this with the cyber stuff as well. There's like people that are just convinced that like AI can't hack anything. When in

41:30Unidentified speaker (UNKNOWN): reality, it's like

41:32Unidentified speaker (speaker_1): a responsibility, right? Okay. In 2018, Elon Musk told me he was gonna fucking sell me uh. Tesla Road so they could do zero to sixty in under two seconds, right?

41:42Unidentified speaker (speaker_3): Right.

41:42Unidentified speaker (speaker_1): And then it was vapor for like eight years.

41:45Unidentified speaker (UNKNOWN): Right.

41:46Unidentified speaker (speaker_1): That and it may still be vapor, but it doesn't mean that I'm not going to like try the next one and see is it vapor? Like it would be crazy for me to be like, oh, it's not possible that this will ever happen because it didn't happen one time in the past. Like that's the

42:03Unidentified speaker (UNKNOWN): most

42:03Unidentified speaker (speaker_1): brain dead view of the world. Like, what? Like, no, we just saw the thing fucking flipped in the air, man. Like, what do you like? I was

42:11Unidentified speaker (speaker_3): about to say, have you seen the latest one? Like

42:13Unidentified speaker (UNKNOWN): the thing

42:13Unidentified speaker (speaker_3): flies, man.

42:14Unidentified speaker (UNKNOWN): This is my point. But like,

42:15Unidentified speaker (speaker_1): but imagine if you're like, No, no, no, that's not real because of twenty eighteen, because I once got fooled by Elon Musk and he told me it was gonna be out at

42:23Unidentified speaker (UNKNOWN): the end. You just

42:24Unidentified speaker (speaker_3): you just have to constantly adjust your priors and like uh um it's

42:28Unidentified speaker (UNKNOWN): I

42:28Unidentified speaker (speaker_3): don't know. I think it it just like with the cyber, especially it is a bit

42:34Unidentified speaker (UNKNOWN): dis

42:34Unidentified speaker (speaker_3): I hate to say this, but it's a bit disappointing because

42:38Unidentified speaker (UNKNOWN): like

42:38Unidentified speaker (speaker_3): DPRK, right, when they're using AI, it's like the most boring what what they're doing with it is so boring. It's like oh they're like um they're like having it like adjust their social engineering scripts, right? So it's like this itty bitty market

42:55Unidentified speaker (UNKNOWN): micro

42:56Unidentified speaker (speaker_1): optimization. But the scary thing though is that like they know what they're doing.

43:01Unidentified speaker (speaker_3): Yeah, yeah. And

43:02Unidentified speaker (UNKNOWN): like it's like

43:03Unidentified speaker (speaker_3): it's worth it's worth tracking, and it's interesting to see. And like, yeah, it is improving their workflows. And you know, we've also seen them do some code stuff, right? So, like, you know, they used to have to code all their front ends and their back ends and their C2s. Now the AI can help. You know, it is what it is. But I think

43:19Unidentified speaker (UNKNOWN): when you've been sold

43:20Unidentified speaker (speaker_3): a story, when you've been sold a story that like the AI can just autonomously hack the world.

43:26Unidentified speaker (UNKNOWN): Right.

43:26Unidentified speaker (speaker_1): And

43:26Unidentified speaker (speaker_3): if DPRK gets it, the world's gonna end. And then you look and it's like, wait, they're having it. Or like the deep fake videos that like actually are worse than the the the prior. priority.

43:37Unidentified speaker (UNKNOWN): The original ones. Yeah. Yeah.

43:39Unidentified speaker (speaker_3): You know what I

43:40Unidentified speaker (UNKNOWN): mean?

43:40Unidentified speaker (speaker_3): It's like it is. It's disappointing. And it's easy for people, I think, to say, oh, like, you know, everyone's just dramatic and sort of write it all off. I think it's for me, the way that I've been approaching it is like one, we we're not seeing the attackers optimize the AI yet. We will, though, 100%. And especially the people that are like AI native, right? It makes

44:04Unidentified speaker (UNKNOWN): more

44:04Unidentified speaker (speaker_3): sense to me that someone that's entrenched in their workflows is optimizing their existing workflows. I think the the step change is gonna be when a new sort of like

44:15Unidentified speaker (speaker_1): another bad country emerges that like

44:19Unidentified speaker (speaker_3): or

44:19Unidentified speaker (UNKNOWN): the younger the

44:20Unidentified speaker (speaker_3): younger the younger DPR H

44:22Unidentified speaker (UNKNOWN): the

44:22Unidentified speaker (speaker_1): younger gen,

44:23Unidentified speaker (UNKNOWN): the younger who

44:24Unidentified speaker (speaker_3): are starting from scratch and everything that you know, instead of optimizing an existing campaign that works really well, we're gonna see them create a campaign AI first, right?

44:34Unidentified speaker (UNKNOWN): But

44:35Unidentified speaker (speaker_0): I think okay, so going back to normies and AI, I think that normies are afraid enough of AI that they do realize that AI is going to hack things. But if you were to ask a normie, do you want the flight tracking system to be written by humans or written by AI? Like 95% of them are gonna say humans. If you said, Do

44:54Unidentified speaker (UNKNOWN): you

44:55Unidentified speaker (speaker_0): want like surgery software? There's about to be a robot that does surgery on the do you want the software to be written by a human or an AI? They're all gonna say human. Do you want like the truck driver? Do you want the truck driver to be a human truck driver or do you want it to be an autonomous truck? And the answer that it's safer for it to be an autonomous truck. Like we're gonna see statistics with cars that show it's safer to have the computer driving. It's going to be safer to have the computer building. And we're in this transition period where there's like this asymmetric computers are attacking, and yes, people are afraid of them, but no one has the confidence to like let computers rip and build the software.

45:35Unidentified speaker (UNKNOWN): Like,

45:36Unidentified speaker (speaker_1): isn't that a sign? Like they're like the fact that like literally AI agents are ripping through our software, like they're our sign.

45:46Unidentified speaker (UNKNOWN): Maybe they're better than us

45:47Unidentified speaker (speaker_1): at this thing, right?

45:49Unidentified speaker (UNKNOWN): The

45:49Unidentified speaker (speaker_0): slop

45:49Unidentified speaker (UNKNOWN): is the human.

45:50Unidentified speaker (speaker_1): If you like if if robots were like destroying humans in the World Cup, right? It'd be weird if you were then like, yeah, but I don't want to drive, like they shouldn't be driving me. Like

46:02Unidentified speaker (UNKNOWN): what?

46:03Unidentified speaker (speaker_1): Like, no, like they're like a much harder thing, a much harder thing in the same category of thing, like controlling the physical world. They're just blowing away our best guys, and you're like, Yeah, I can't drive better than me, man. Shut the fuck up.

46:17Unidentified speaker (UNKNOWN): Normies think

46:18Unidentified speaker (speaker_0): this way. I'm telling you, normies think this

46:19Unidentified speaker (UNKNOWN): way.

46:20Unidentified speaker (speaker_1): So, Austin, I I have a question for you. Your job is builder enablement. Um, maybe you should change it to like builder. save enablement or something. Because we need we need to we need to stop people from getting wrecked, right? When they're building shit. So, how much of your work at the moment is like getting people to understand that the benefits of like red teaming and using agents and and formal verification and all of that stuff? Like, you know, I know like ETH skills, right? Was fucking amazing. I remember the first time that I did my red team thing.

46:54Unidentified speaker (UNKNOWN): I

46:54Unidentified speaker (speaker_1): gave vanilla Claude Eat Skills, and suddenly it was like the, you know, I know Kung Fu guy. It was like, oh yeah, I'm just spinning up Foundry and whatever. So like that stuff is amazing. Um, those skill files are amazing. Um, but like the state of the art has obviously moved on. So, like, how much time like, and I hope the answer is 100%. of the time, but like, how much how much of your time is like focused on making sure that you're giving uh Ethereum engineers like the best tools to be defending themselves, etc. Like, what what's the

47:28Unidentified speaker (UNKNOWN): it's not

47:28Unidentified speaker (speaker_0): a hundred percent? Yeah, I I wish it was, but it's not. Yeah. I think like a lot of it is about like even the mimetic value of one dollar audit, like having one dollar audit at as this meme that like there's so many people that shit on it and say you're getting what you're paying for, but it it's meant to prove that like the game has completely changed and you should be using AI in your pipeline to audit your contract. And it should be more of like a CI C D thing rather than a twenty thousand dollar audit. But then like on the other end, we're still telling everybody you should have that twenty thousand dollar audit. Like there's there's

48:04Unidentified speaker (speaker_1): when does that shift? Like, realistically, when does it shift? Because it is an impediment to people, right? Um, and you know, as someone who's seen a lot of fucking audits,

48:15Unidentified speaker (speaker_0): yeah,

48:16Unidentified speaker (speaker_1): like humans are not infallible. An audit is like, and there is some level of like just brute forcingness to this problem of like get enough eyeballs to look. at a contract that has like

48:29Unidentified speaker (UNKNOWN): you

48:29Unidentified speaker (speaker_1): know

48:30Unidentified speaker (UNKNOWN): a

48:30Unidentified speaker (speaker_0): lot of people looked at the balancer contract right like yeah but it took

48:35Unidentified speaker (UNKNOWN): like

48:35Unidentified speaker (speaker_1): AI

48:36Unidentified speaker (speaker_0): frontier level AI

48:38Unidentified speaker (UNKNOWN): to

48:38Unidentified speaker (speaker_0): find those those edge cases yeah

48:41Unidentified speaker (speaker_1): so

48:43Unidentified speaker (UNKNOWN): so when

48:44Unidentified speaker (speaker_1): don't

48:44Unidentified speaker (UNKNOWN): you

48:45Unidentified speaker (speaker_0): yeah no I mean

48:47Unidentified speaker (UNKNOWN): I

48:47Unidentified speaker (speaker_0): just got mythos access just now and I'm paying API rates for it. So I'm barely having it like it's working on my hardware wallet and other things.

48:55Unidentified speaker (UNKNOWN): Yeah.

48:55Unidentified speaker (speaker_0): So like

48:57Unidentified speaker (UNKNOWN): I

48:57Unidentified speaker (speaker_0): think it has to do with the models, and I don't think that they're at the point where you can just point mythos at a smart contract and it can do better than an audit agency, because an audit agency is going to be a bunch of humans that have done this before, plus all their tools, plus mythos, and they're gonna use all of that. And I think that it it's gonna be a while before one dollar audit is you know on par with a $50,000 audit.

49:21Unidentified speaker (speaker_1): I mean, the the audit thing was always like there are maybe 50 engineers who have the right mindset, right? And and also the cyclical nature of crypto, right? Like, you know, I would see a new audit firm pops up, they're amazing,

49:35Unidentified speaker (UNKNOWN): they

49:35Unidentified speaker (speaker_1): make a million dollars like in you know 2017. twenty seventeen during the ICO boom or two million or five million or whatever the fuck, you know. They started charging like a hundred thousand dollars an audit or whatever for an ERC twenty contract. Like they make a fuckload of money and then there's a bust and then they go back to the real world. And they like usually don't come back, right? And so, you know, there's not this consistency of like audit teams. Like there's very few teams that I know that have like stuck it out. I mean, you know, obviously see consensus has audits and uh um

50:03Unidentified speaker (UNKNOWN): yeah,

50:03Unidentified speaker (speaker_1): I remember them pitching me formal verification like 2019. Uh

50:07Unidentified speaker (speaker_0): verification is back, baby. It's it's

50:09Unidentified speaker (UNKNOWN): I'm

50:09Unidentified speaker (speaker_0): like we're talking about

50:11Unidentified speaker (speaker_3): Lubin is he's like I already know this. We've been doing this forever. It's been it's been funny. When you're

50:20Unidentified speaker (UNKNOWN): talking about

50:20Unidentified speaker (speaker_0): formally verified clients now, and then that's super

50:23Unidentified speaker (UNKNOWN): hard with

50:23Unidentified speaker (speaker_0): all the networking layer and everything, but they're talking about instead of six different clients, and if one breaks we're fine, it's one client that's formally verified, which is really cool.

50:33Unidentified speaker (UNKNOWN): We

50:33Unidentified speaker (speaker_1): need to formally verify the frequencies, man. That's

50:35Unidentified speaker (speaker_0): uh why didn't they formally verify the frequencies?

50:39Unidentified speaker (speaker_1): So so uh so um I think yeah like to go back to for for like small teams up and coming teams etc et cetera, right like the one $1 order is a meme. Um, but like continue. So this week, um, actually for like the last like I want to say two weeks, there's been a guy or a set of guys that's running agents. Um, I suspect it's Deep Sea because I've looked at some of the traces and I I know that guy well enough. Um, that has been literally trying to attack my accounts, like me. personally, like systematically going through and like trying to attack my accounts. And um we have a continuous red team exercise at this point where like it just runs all the time. And so we pick this up. We've got agents that are like watching the traces and looking at logs and everything. We pick this up and we're like kind of just watching it just to see because there is some element of like, you want to what you know, if you've got a burglar out. like and

51:39again this is this is like probably my insanity, but you know, you got a burglar outside of your house. I look at that as an opportunity because this guy's going to test your defenses, right? And oh

51:52Unidentified speaker (UNKNOWN): yeah, no, it is

51:53Unidentified speaker (speaker_1): a shotgun sitting on your lap, right? And you're like,

51:56Unidentified speaker (UNKNOWN): all right,

51:57Unidentified speaker (speaker_1): bro, like let's

51:58Unidentified speaker (UNKNOWN): see. You know, and then

51:59Unidentified speaker (speaker_1): and then then the burglar finds like the little gap in the fucking window that you didn't realize you had, and like fucking the window's gone, and then you blow his brains out. Um but like

52:10Unidentified speaker (speaker_3): and then

52:10Unidentified speaker (UNKNOWN): and then

52:10Unidentified speaker (speaker_3): you we and then you fix the window. So that

52:13Unidentified speaker (UNKNOWN): knowing there's gonna be a hundred more

52:14Unidentified speaker (speaker_0): burglars coming.

52:15Unidentified speaker (speaker_1): Yeah, exactly.

52:17Unidentified speaker (UNKNOWN): No, and

52:17Unidentified speaker (speaker_3): and this is so

52:18Unidentified speaker (UNKNOWN): for

52:19Unidentified speaker (speaker_3): teens for let's see, over the past I would say six months or so, my advice has gone from like my generic what's the one thing you could do to crypto teams has shifted from um some of the like more classic advice to like right now. My advice is like monitoring. Like you need to know, you need to see, you need to have visibility, and you need to just crank that through the roof. Um, and there's two reasons for this. One is that there is just a lot more activity that's happening and like kind of low brow activity, right? Like things people agents are just like poking at something.

52:58Unidentified speaker (UNKNOWN): They can stumble on

52:59Unidentified speaker (speaker_1): they can stumble on something that

53:01Unidentified speaker (UNKNOWN): like

53:01Unidentified speaker (speaker_1): no one else

53:02Unidentified speaker (UNKNOWN): they

53:02Unidentified speaker (speaker_3): wonder

53:03Unidentified speaker (UNKNOWN): can.

53:04Unidentified speaker (speaker_3): Also like uh 50 agents can stumble on something, get access to something, and if you don't know it, um because they don't manage to do the next hop that takes the money or does something damaging you know it just is again it's like

53:20Unidentified speaker (UNKNOWN): the

53:21Unidentified speaker (speaker_3): 50 agents go wander it through the door and they're like ah whatever but that fifth the 51st agent who also has a human looking

53:27Unidentified speaker (speaker_1): yeah

53:28Unidentified speaker (speaker_3): or like is able to you know iterate or whatever um so yeah if you have good monitoring if you have good detection if you have good eyes then it's so much easier for you to be constantly sort of patched patching these things and seeing where the attacks are. Um that's been that's something historically that I just haven't really it has been a priority for most teams, right? Like that's

53:50Unidentified speaker (UNKNOWN): red teaming your web

53:51Unidentified speaker (speaker_0): two. Your your web two

53:52Unidentified speaker (UNKNOWN): stuff

53:52Unidentified speaker (speaker_0): should be red

53:53Unidentified speaker (UNKNOWN): teamed like crazy.

53:54Unidentified speaker (speaker_0): And I

53:55Unidentified speaker (UNKNOWN): think that like

53:55Unidentified speaker (speaker_0): we in the in the past we just hired a very smart person to handle the web 2 two stuff and they just know what to do and what not to do. Even though that person has been there and you've gone through everything, send the red team in and be constantly

54:08Unidentified speaker (UNKNOWN): banging

54:09Unidentified speaker (speaker_0): on your own shit.

54:10Unidentified speaker (speaker_3): Yeah. Um, and then the other thing is that it's the AI is very good at leveling your own team up. You can have a very small team, and the AI can help you actually effectively monitor all your different systems and like catch an omelet. anomalies and stuff in ways that used to be used to have to have like really skilled people, right? And set it up and right. So it's not enough to just like have the eyes. You have to use the eyes correctly

54:37Unidentified speaker (UNKNOWN): to be able to

54:38Unidentified speaker (speaker_3): detect. But

54:39Unidentified speaker (UNKNOWN): the AI is very

54:40Unidentified speaker (speaker_3): good at this, right? The AI is so good at being like, that's anomalous. Right. What is this? Right. And that's what's what

54:47Unidentified speaker (UNKNOWN): everyone has logs. all about.

54:48Unidentified speaker (speaker_1): But no one reads them, right? Like

54:50Unidentified speaker (UNKNOWN): the problem. process

54:51Unidentified speaker (speaker_1): wasn't logging, right? Like everyone had like there's logs everywhere, right?

54:54Unidentified speaker (UNKNOWN): But like

54:54Unidentified speaker (speaker_1): you gotta

54:55Unidentified speaker (speaker_3): be surprised.

54:56Unidentified speaker (UNKNOWN): You

54:57Unidentified speaker (speaker_3): would be surprised.

54:58Unidentified speaker (UNKNOWN): It's so cool how an agent.

55:00Unidentified speaker (speaker_0): An agent can just like get into your like I I am like DevOps, right? I'm not a security engineer, I'm a DevOps engineer. I know how to like set the boxes up and make sure most ports are closed and all that. But it's so cool. You can just send an agent in and he's like, Yeah, dude, there's like

55:15Unidentified speaker (UNKNOWN): this slow

55:15Unidentified speaker (speaker_0): traffic from Hong Kong that's been banging on all your shit all week. Like you, and and he's like, you know, it's it's comes down to like your SSH key and you know, e-pair cryptography, so you're good. And I'm like, Oh, maybe not, man.

55:31Unidentified speaker (speaker_1): That guy has

55:32Unidentified speaker (UNKNOWN): me reading the news.

55:33Unidentified speaker (speaker_0): Yeah, exactly.

55:35Unidentified speaker (speaker_3): Yeah, exactly. So

55:36Unidentified speaker (UNKNOWN): it's I

55:37Unidentified speaker (speaker_3): don't know, that's it's so interesting. It's and it really, I do feel like the next six months are gonna be wild in this regard.

55:46Unidentified speaker (speaker_1): So so um Austin, then like concretely for for a young team that's working on stuff, right? Like best

55:56Unidentified speaker (speaker_0): practices, like

55:57Unidentified speaker (UNKNOWN): you

55:57Unidentified speaker (speaker_0): you should know best practices for your smart contracts, you should be using like pre vetted open Zeppelin stuff as much as you can. You should be building things on top of things that are already audited. You should be doing tons of like your own AI audits. You should write hella tests, hella fuzzers, all of this, the obvious stuff that you should be doing. You need to like 10x and have it ready. And then still you go to the audit team. But maybe there was a more granular question there.

56:23Unidentified speaker (speaker_1): Well, I mean, yeah, like

56:25Unidentified speaker (UNKNOWN): I

56:25Unidentified speaker (speaker_1): the amount of teams that I I've spoken to, they're like, oh, audits, like, you know, it's such a big cost. We're changing our code, then we're gonna audit it again. Uh, you know, it is it's like it's a huge impediment, I would say, right? So

56:40Unidentified speaker (speaker_3): okay, so this is this is why I keep this is why I think my advice has been shifting towards like more monitoring and detection. Okay, so imagine you have like a smart contract, like, yeah, every time you change it, every time the world changes, you're gonna, you're re-auditing, right? Which has a cost, which also it's it's very fallible. Like it just is. However, if you're able to um especially like with AI, it's so much easier, it's so much faster, it's so much more robust. You don't need to have like really advanced engineers and and understanding of security systems to do this. The AI can do this pretty damn well Right. If you can have uh your systems designed in such a way or upgraded in such a way where uh even if something bad happens, it's monitored and it's detected in instantly and then it's reacted to autonomously, you're still gonna be fine. Right. And that's that layer that historically in crypto we just have not had. But imagine like okay, with your smart contract, you can audit it again and again and again. And basically you're betting on the strength of that audit. Or what if you say we're gonna audit it and we're gonna expect the audit covers whatever, like 90%? It gets us 90% of the way there. But the other 10%, we're gonna have an additional like circuit breaker layer

58:00or an additional rug layer, whatever. some sort of layer where if XYZ happens, it's just gonna pull the plug

58:09Unidentified speaker (UNKNOWN): and

58:09Unidentified speaker (speaker_3): that's gonna save

58:11Unidentified speaker (UNKNOWN): the

58:11Unidentified speaker (speaker_3): money. And that's gonna cover that that long tail 10% of all the other incidents that you know

58:18Unidentified speaker (UNKNOWN): no human or

58:19Unidentified speaker (speaker_3): AI would find and you know, whatever. And that's so like that's phenomenal. And AI can really like level up. your own team to be able to do that. Because historically, again, you the team has to have the the knowledge and the

58:34Unidentified speaker (UNKNOWN): skills.

58:35Unidentified speaker (speaker_1): Historically, you were asleep. You were like doing the brain thing where you switch your brain off for eight hours every day.

58:41Unidentified speaker (speaker_3): Yeah.

58:42Unidentified speaker (speaker_1): And the hackers knew that, and they would attack you at the time when you're asleep every

58:47Unidentified speaker (UNKNOWN): time.

58:47Unidentified speaker (speaker_1): Right. So now, so so sorry, just like to be really clear, what you're suggesting is that you can um set up all of your logging, then have like a continuous monitoring system that's run with an agent that has the ability to like kill switch stuff,

59:04Unidentified speaker (speaker_3): or yeah, I mean, yeah, you can 100%. Or

59:07Unidentified speaker (UNKNOWN): you

59:08Unidentified speaker (speaker_3): can have different layers, right? So it can throw alerts, it can throw pager duties or whatever. That's still a an alerting system, right? But if you set say, just like for simplicity's sake, right? Let's say that you have a smart contract, the smart contract holds money, right? You can say under no circumstances should XYZ happen, under no circumstances should $200 million ever leave this contract, right? Right. Then you can you can have the AI, and this is like where the AI really like levels up your own team, you can have the AI iterate and find and design those,

59:40Unidentified speaker (UNKNOWN): you

59:40Unidentified speaker (speaker_3): know, the most effective

59:41Unidentified speaker (UNKNOWN): sort of it

59:43Unidentified speaker (speaker_0): shouldn't be a guy from the outside flipping a switch. It should be like in Solidity if 200

59:47Unidentified speaker (UNKNOWN): million

59:47Unidentified speaker (speaker_0): moves here, like

59:48Unidentified speaker (UNKNOWN): flip this switch

59:49Unidentified speaker (speaker_0): and don't like

59:49Unidentified speaker (UNKNOWN): interesting.

59:50Unidentified speaker (speaker_0): Yeah. Yeah.

59:51Unidentified speaker (UNKNOWN): And

59:51Unidentified speaker (speaker_3): then yeah, because the first step is figuring out what are like the things that you don't want to happen, right? You have to define them. And then the second one is figuring out how to do that safely without creating more risk. Probably it's so much more accessible these days, though, right? Because

1:00:05Unidentified speaker (UNKNOWN): again,

1:00:06Unidentified speaker (speaker_3): historically, you have a few dudes creating the most awesome smart contract. And then they're like, okay, but we need to mitigate the risk that it might be insecure. So you get audits. The audits don't get you 100% of the way. So you do more audits. Like it's just not a sustainable pattern, right? The audits get you a certain percentage of the way. And then for that long tail of risk, you have. again like systems that magically prevent total

1:00:32Unidentified speaker (UNKNOWN): time. This is

1:00:32Unidentified speaker (speaker_1): a controversial uh thing, right? But like, you know, audits catch a lot of stuff, but they never catch the thing that kills you.

1:00:42Unidentified speaker (speaker_3): Yeah.

1:00:42Unidentified speaker (UNKNOWN): If

1:00:43Unidentified speaker (speaker_1): you're dead, it's because an audit missed it.

1:00:45Unidentified speaker (UNKNOWN): Like

1:00:46Unidentified speaker (speaker_1): realistically, right? Like, how many times does fucking balancer audit, order like you know

1:00:50Unidentified speaker (UNKNOWN): Exactly?

1:00:51Unidentified speaker (speaker_3): Exactly. You know,

1:00:52Unidentified speaker (UNKNOWN): like

1:00:53Unidentified speaker (speaker_3): there's

1:00:54Unidentified speaker (UNKNOWN): not,

1:00:54Unidentified speaker (speaker_3): you I don't know. I just I have very little faith. Like formal verification, it's like we're just we're playing with like the that that 1%, 1%, 1%. You know, it's like okay, the audit gets us 90%, the formal verification will get us another five percent. So now you but you still have five percent that but it might just completely collapse. And

1:01:14Unidentified speaker (UNKNOWN): at the AI audit

1:01:14Unidentified speaker (speaker_0): auditor has nothing to to lose either. The AI auditor, you're like, dude, I just got hacked. Well, you're absolutely right. I see that

1:01:23Unidentified speaker (speaker_3): sounds

1:01:23Unidentified speaker (UNKNOWN): like

1:01:23Unidentified speaker (speaker_0): lighter's guardian mode froze withdrawals and they so they had a circuit breaker and five million back. Yeah.

1:01:31Unidentified speaker (UNKNOWN): Yeah. So

1:01:31Unidentified speaker (speaker_1): just put a circuit

1:01:32Unidentified speaker (UNKNOWN): breaker.

1:01:33Unidentified speaker (speaker_3): That They they did get money back. Or their circuit breaker did work.

1:01:37Unidentified speaker (UNKNOWN): I

1:01:37Unidentified speaker (speaker_0): think so. That's

1:01:39Unidentified speaker (speaker_3): amazing. This is

1:01:41Unidentified speaker (UNKNOWN): what I

1:01:41Unidentified speaker (speaker_3): mean. This is what I mean. Like, you know,

1:01:44Unidentified speaker (UNKNOWN): and by the way,

1:01:45Unidentified speaker (speaker_3): mitigating the risk, right? Um, meaning that like instead of being wrecked a hundred percent of the way, you can be wrecked five percent of the way. That's that's also very nice, right?

1:01:57Unidentified speaker (UNKNOWN): While still

1:01:58Unidentified speaker (speaker_0): being permissionless finance, right? Like there's

1:02:00Unidentified speaker (UNKNOWN): exactly it there should

1:02:00Unidentified speaker (speaker_0): not be a guy that can be like, no. Yeah,

1:02:04Unidentified speaker (UNKNOWN): maybe

1:02:04Unidentified speaker (speaker_0): there should be, but you don't want that.

1:02:06Unidentified speaker (speaker_1): So I think maybe, maybe just like to restate this. Um we used to have this hesitancy to have like external systems that could shut things down or whatever, right? Because then, oh, what if someone comes in and shuts it down? Right. And that was like a lot of the reason why we didn't have kill switches and things like that.

1:02:27Unidentified speaker (speaker_3): Yep.

1:02:27Unidentified speaker (speaker_1): And then I think we kind of solve that to some extent with like, you know, like multiple gates, right? So anyone can switch a thing off. Like, would you rather Ave is off for an hour or a smoking crater? Like, obviously, like one is preferable. If you can't tell the difference between those two,

1:02:49Unidentified speaker (UNKNOWN): like

1:02:50Unidentified speaker (speaker_1): seriously. fucking take a look in the mirror, man. Like, but there are people back in the day that would have said, I prefer the smoking crater, because if Ave can be turned off, then I'm out.

1:03:01Unidentified speaker (speaker_0): Ethereum Classic, guys. Now

1:03:03Unidentified speaker (speaker_1): we're a bank.

1:03:03Unidentified speaker (speaker_3): Yeah. Right.

1:03:04Unidentified speaker (UNKNOWN): And so

1:03:05Unidentified speaker (speaker_1): Ethereum Classic. Fucking thanks, Mary. Um, but yeah, so uh, so I just I think that like the design space of ensuring that we've got like systems that can, you know, have kill switches and stuff like we've sort of accepted at this point that that is um far preferable to an immutable smart contract uh that you know has no escape hatch at

1:03:30Unidentified speaker (UNKNOWN): all.

1:03:31Unidentified speaker (speaker_3): Yeah. Well, and it's also there's a lot, there's just so much room. It's not black and white, right? There's so much gray area where being able to have like a detection and a system that prevents total loss, right? Like catastrophic failure doesn't necessarily mean put one dude in charge of the EOA that can kill the system, right? Like, um, just like as an example, like if something's trying to do, if if XYZ happens, then put it in like withdrawal only. mode for existing users or for right. So like it's like lock down this edge or whatever.

1:04:09Unidentified speaker (UNKNOWN): A circuit breaker

1:04:10Unidentified speaker (speaker_0): can still be immutable code, right? You can have

1:04:14Unidentified speaker (UNKNOWN): the the

1:04:15Unidentified speaker (speaker_0): protocol,

1:04:15Unidentified speaker (UNKNOWN): and then

1:04:16Unidentified speaker (speaker_0): you can have the circuit breaker, and no human can change any of that. And still, if 200 million tries to go out the door, it flips something. And you

1:04:24Unidentified speaker (UNKNOWN): have it, you know, like

1:04:25Unidentified speaker (speaker_1): we've seen this movie before. It's called the fucking DAO hack. And then you sit there for 28 days and you're like, Oh, okay. Like, so at some point someone needs to be able to like upgrade shit, right? Like you and you know, I mean

1:04:38Unidentified speaker (speaker_3): proxy card. Yeah, and that's the that's the

1:04:41Unidentified speaker (UNKNOWN): but that was good,

1:04:41Unidentified speaker (speaker_0): right? Like that was a saving grace of the DAO, the fact that it took 30 days because they were able to like get everything together and in White Hat Hatted. it. Right.

1:04:48Unidentified speaker (UNKNOWN): For

1:04:48Unidentified speaker (speaker_1): sure.

1:04:48Unidentified speaker (speaker_3): Yeah, yeah.

1:04:49Unidentified speaker (speaker_0): If exactly

1:04:49Unidentified speaker (UNKNOWN): if it

1:04:49Unidentified speaker (speaker_0): would have spread everywhere. it would have

1:04:51Unidentified speaker (UNKNOWN): Yeah. It

1:04:52Unidentified speaker (speaker_3): would have been over. Yeah, exactly.

1:04:54Unidentified speaker (speaker_1): But the the flip side, the sorry, the flip side, the argument I was I was sort of making is yes, it was good that we had 28 days to like or seven days. It was like seven days, then like there's a whole chain of gates, right? Um back in in prehistory. Um, but uh but like if it was just a standard proxy contract, right, that had that time lock, then you could have just gone in and like fixed the bug.

1:05:18Unidentified speaker (speaker_3): Fixed the bug

1:05:18Unidentified speaker (UNKNOWN): in

1:05:18Unidentified speaker (speaker_3): the first place. Yeah.

1:05:19Unidentified speaker (UNKNOWN): But you

1:05:19Unidentified speaker (speaker_1): couldn't fix the bug, right? You had you were racing against time to fix the chain, not the bugs. That was the issue, right? Like, you know, not everyone has access to the chain. Maybe Austin, uh, next time something bad happens, we'll just give you a call and you can help us.

1:05:35Unidentified speaker (UNKNOWN): I've

1:05:35Unidentified speaker (speaker_0): I've got a validator,

1:05:36Unidentified speaker (UNKNOWN): I've got

1:05:37Unidentified speaker (speaker_0): two validators here that have a haven't validated validated a single block for a hundred days. So I'm a dry spell.

1:05:42Unidentified speaker (UNKNOWN): I'm

1:05:42Unidentified speaker (speaker_0): I'm in

1:05:43Unidentified speaker (UNKNOWN): a

1:05:43Unidentified speaker (speaker_0): dry spell. I can

1:05:44Unidentified speaker (UNKNOWN): I can

1:05:47Unidentified speaker (speaker_1): All right. Let's wrap it up there. Thank you very much, everyone. See you next week. Um thanks for joining us. Um and stay safe out there. Nothing you hear on uneasy money is financial advice. We're just three builders talking about what's happening on chain. and we want you to always do your own research before aping in. You can find all our disclosures at unchaincrypto.com slash uneasymone.

This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights