Last week an unusual meeting took place in Washington DC. US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned banking leaders to discuss what could mark a new era of cybersecurity.
Anthropic has been in the news of late. It recently had a dispute with the US Department of Defense. Today, however, the issue is not with Claude AI, but with Mythos Preview.
Why is Anthropic’s Mythos Gaining Attention?
According to Anthropic, Mythos Preview is a new general-purpose language model. While it performs strongly across the board, it is strikingly capable at computer security tasks. In response, Anthropic launched Project Glasswing.
The decision was driven by internal testing in which the company’s team discovered that Mythos Preview can identify and exploit zero-day vulnerabilities in every major web browser when instructed to do so. These vulnerabilities are often extremely difficult to detect.
Polymarket commented on X that “Claude Mythos is reportedly intelligent enough to ‘spot weaknesses in almost every computer on earth’”.
Anthropic’s team made it clear that they did not explicitly train Mythos Preview to have these capabilities. Instead, they “emerged”.
The long-term outlook from Anthropic is optimistic: the company believes defense capabilities will eventually dominate, leading to a more secure world with better-hardened software, which is partly thanks to code written by these very models. The transitional period, however, will be “fraught.” That is why immediate action is required.
“We believe it is once again time to launch an aggressive forward-looking initiative,” the statement concludes. “But this time, the threat is not hypothetical. Advanced language models are here.”
What is Project Glasswing?
Project Glasswing is Anthropic’s direct response to that transitional risk. It gives a limited group of critical industry partners and open-source developers early access to Mythos Preview. The goal is to enable defenders to begin securing the most important systems before models with similar capabilities become widely available.
Dario Amodei, CEO of Anthropic, shared on X: “I’m proud that so many of the world’s leading companies have joined us for Project Glasswing to confront the cyber threat posed by increasingly capable AI systems head-on.”
These firms include AWS, Apple, Google, Microsoft, NVIDIA, JPMorgan Chase and others.
Banks Should Worry
The banking industry is already navigating significant disruption. While there has been innovation, many institutions have struggled to keep pace with the relentless need to invest in cybersecurity.
Last Tuesday, Bessent and Powell gathered Wall Street leaders at the Treasury to discuss the risks raised by Anthropic’s Mythos Preview. Attendees included leaders from systemically important banks such as Bank of America, Morgan Stanley, Goldman Sachs, Citigroup, JP Morgan Chase, and others.
The meeting was particularly significant as Powell prefers to have an independent role. He wouldn’t normally collaborate so closely with the Treasury. However, the concerns about this new threat were so big that it could be referred to as systemic risk, Bloomberg reported.
Bessent had already spoken to tech CEOs together with Vice President JD Vance before Mythos Preview was launched. That discussion focused on how the private sector should respond to the coming wave of AI-powered cyber threats.
If banks aren’t yet worried, it is now very clear that the Treasury, the White House and the Federal Reserve are.
Mythos Sparks Global Response
The new revelations from Anthropic haven’t just caused concern in Washington. In Europe, Germany’s Federal Office for Information Security (BSI) has been in direct contact with the company.
“We are in contact with the manufacturer Anthropic with respect to Claude Mythos,” announced BSI President Claudia Plattner. She added that BSI was taking Anthropic’s announcements “very seriously” and that the development “raises questions of national and European security and sovereignty.”
The Bank of England is also moving quickly. It is planning a high-level discussion on Mythos within the next two weeks that will include representatives from the Treasury, the Financial Conduct Authority and the National Cyber Security Centre.
The message is unmistakable: whether in Washington, Berlin or London, regulators and governments now view advanced AI models like Mythos Preview as a live national-security issue. For the global banking system, one which runs on complex, often legacy infrastructure, there is no time to waste.
The age of AI-native cybersecurity has arrived. The only question left is which institutions will treat it as a crisis, and which will treat it as the biggest competitive opportunity in a generation. Either way, we hope we will share these stories with you here at DisruptionBanking.
Author: Andy Samu
