Episode summary
What the Fed Can Do About Anthropic’s Latest System
Wall Street Week
· · Hosts: Chrystia Freeland, David Westin · Guests: Scott Bessent, Jay Powell
Summary
Anthropic’s latest system, Mythos, is described as too dangerous for public release because it can autonomously probe bank software, find vulnerabilities, and act at high speed across systems. The host says Treasury Secretary Scott Bessent and Fed Chair Jay Powell summoned major bank leaders, and the discussion centers on whether the financial system is prepared for an AI tool that can outpace current defenses.
An analyst argues the main risk is systemic: a sufficiently capable model could expose weaknesses in bank software and enable cyberattacks that disable individual institutions and, potentially, the broader banking system. The analyst says regulators are not ready for this jump in threat level, but they can still ask better questions, facilitate coordination, and help banks plan for containment if an attack occurs.
The open-source angle is more optimistic. A guest from Hugging Face says banks may not need Mythos itself to build better defenses, since existing tools, third-party testing, and shared expertise could help create stronger safeguards. But the guest also says full autonomy is the wrong design choice for safety, and that semi-autonomous systems with guardrails and human oversight are better suited to finance.
Key takeaways
- The host says Anthropic’s Mythos is being held back from public release because it is too dangerous, even as regulators brief major banks on the risk.
- An analyst says the biggest concern is a cyberattack that could disable individual banks and potentially the broader banking system.
- The analyst says the Fed cannot stop a serious attack outright, but it can help contain damage through liquidity support, coordination, and contingency planning.
- A guest from Hugging Face says banks can use open-source tools, third-party testing, and existing cybersecurity practices to strengthen defenses without relying on Mythos itself.
- The guest argues semi-autonomous AI is safer than full autonomy because finance needs clear context limits, accountability, and human oversight.
Notable quotes
“that its new mythos preview is too dangerous for public release”
“there are real concerns that this new much more sophisticated form of AI could detect and likely will detect very meaningful vulnerabilities in banks software”
“the government can play a very meaningful role helping to contain the damage”
“-Mythos is really on the cutting edge of cybersecurity right now”
“-Semi-autonomous. I've actually written papers on this and op eds and I'm very opinionated about this, actually”
How the episode opens
Anthropic has said that its new mythos preview is too dangerous for public release. And Secretary Bessent and Chair Powell called in the heads of the major banks to warn them about it. -I applaud Secretary Bessent for doing that.
One of the things we always were focused on the risk to the banking system, the risk for any kind of cyber-attack. You're just getting in and change the ones and zeros around in the banking system. So much of it rests on our banking system.
So nothing is more important. -We need to do something about it. I don't think there is a framework.
You know everybody is keen to have a framework within which to operate. I don't think there is a governance framework that is there to actually mind those things. We need we need to work on that.
Westin: Kathryn Judge is a professor at Columbia Law School specializing in the Fed. She explained the risks that mythos could pose for the banking system and what regulators can do to manage them. -The big deal is there are real concerns that this new much more sophisticated form of AI could detect and likely will detect very meaningful vulnerabilities in banks software in ways that could expose them to cyber-attacks that would disable individual banks and potentially the broader banking system.
-There is the danger of bad actors getting access to it. Is there a flip side to it also? They may give the banking system and banks some defensive tools although as might lack.
-Yeah, I mean, this is always the interesting question is like AI fighting AI and what does that look like. And it's also a question of what that looks like for the structure of the banking system. Again the U.S.
has always had a very diffuse banking system.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai cybersecurity
- fed regulation
- autonomous agents
- banking software vulnerabilities
- open source tools
Companies and tickers mentioned
- Anthropic
Watch the episode
Watch “What the Fed Can Do About Anthropic’s Latest System” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Wall Street Week
-
What American Manufacturers Are Telling the Fed about the Economy -
Europe’s Welfare State Faces a Reckoning -
How the Blue Wave Could Reshape America’s Economy -
Inside Alaska’s Oil Paradox -
Wall Street Week | Blue Wave Economic Impacts, European Welfare Woes -
The $40 Trillion Problem Washington Won’t Talk About
Wall Street Week, this episode and its artwork belong to Bloomberg. Octagon is not affiliated with the show; this page is an independent summary.
