Episode summary
Dylan Jovine | Anthropic’s New AI Model Is Being Kept Secret: SaaS Apocalypse Explained
Summary
Dylan Jovine argues that Anthropic’s withheld model could materially change cybersecurity because, during testing, it reportedly identified and exploited zero-day vulnerabilities across every major operating system and web browser. He says the model’s ability to uncover thousands of previously unknown flaws, including some dating back decades, suggests AI can do in minutes what once took human teams far longer.
He says the story looks credible because Treasury Secretary Scott Bessent and Fed Chair Jerome Powell reportedly pulled in leaders from Apple, Google, and major banks to discuss it. On that basis, he describes Anthropic’s Project Glasswing as a way to let critical-infrastructure companies test systems before a public release, from banks and software firms to water utilities, hospitals, and national-security targets.
The broader investment implication, in his view, is pressure on cybersecurity and traditional software firms. He says AI is already helping his own business replace outside vendors and cut costs, and he frames that as part of a larger “SaaS apocalypse” in which software-as-a-service businesses may be disrupted as AI absorbs more of their functions.
Key takeaways
- Dylan Jovine says Anthropic’s new model was held back because it could identify and exploit zero-day vulnerabilities in major operating systems and browsers.
- He says the reported discovery of thousands of unknown flaws, some decades old, shows AI can search code faster than human teams.
- Jovine says the involvement of Scott Bessent and Jerome Powell makes the Anthropic warning sound credible, not just promotional.
- He says Project Glasswing could help critical infrastructure firms test systems before a dangerous model is public.
- Jovine argues the “SaaS apocalypse” is real, with AI threatening cybersecurity and other software businesses while lowering his own operating costs.
Notable quotes
“During our testing, we found that mythos, the name of the new model, is capable of identifying and then exploiting zeroday vulnerabilities”
“it's going to work with tech firms and other major systems core systems critical infrastructure whatever banks software firms Microsoft Apple”
“If one model of AI took reportedly by one count, eight minutes to find security vulnerabilities”
“we're using AI to do a lot of things that we used to have to pay other vendors for”
“the software as a service apocalypse. basically that AI has the potential to really eat the lunch of cyber security firms”
How the episode opens
Good morning. This is Dylan Joven with Behind the Markets. Happy Tuesday.
Today's Tuesday, April 14th. Gosh, there is so much to talk about. Where do I begin?
I'm actually going to start with Anthropic. Last week, Anthropic released a note saying that their new AI model is too dangerous to release to the public. During their testing, they said, quote, "During our testing, we found that mythos, the name of the new model, is capable of identifying and then exploiting zeroday vulnerabilities in every major operating system and every major web browser when directed by a user to do so.
End quote. In plain English, Anthropic is saying that its newest AI model, which is not being released, they're holding it back, has found security holes in basically the major systems that power well everything. Computers, banks, stock exchange, everything, Apple, Microsoft, Bank of America.
It's crazy. And you can't release a model like this because bad actors are going to use it to crash the stock exchange, steal money from banks, crash all of our iPhones, crash your computer. And again, what it says basically is that they found thousands of security vulnerabilities that were previously unknown to software developers.
And some of these vulnerabilities date back more than two decades. And I always imagine Microsoft for example, they build their core code. We use Office now.
Maybe their code's built on top of other code and on top of other code and on top of other code. And there's some code or banks banks have code that is built many decades ago and building on top of a lot of this code. You know what I mean?
And I imagine that this AI, not that I imagine, this is exactly what happened. They actually say to this AI, "Okay, go test this browser, Chrome browser, or go test this bank's website and find vulnerabilities.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai model secrecy
- zero-day vulnerabilities
- cybersecurity impact
- project glasswing
- saas disruption
Companies and tickers mentioned
- AAPL
- GOOGL
- GOOG
- Anthropic
- Apple
Watch the episode
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Behind The Markets Podcast
-
Nvidia CEO Out: 4 Reasons I’m Not Panicking About Nvidia -
The Mag 7 Sell-Off: My 3 Secrets to Picking Stocks -
Morgan Stanley’s AI Winners List Is Full of Bank Stocks -
Anthropic IPO: Why I’m Not Rushing to Buy at $1 Trillion -
Chip Stocks Sell-Off: Why the AI Build-Out Just Went From 3 to 5 Years -
Midterm Election Results: The Stock Market’s Winners and Losers
Behind The Markets Podcast, this episode and its artwork belong to their owners. Octagon is not affiliated with the show; this page is an independent summary.
