“Methos”… the digital monster that terrified banks | technology

While the world looks to artificial intelligence as a tool to improve productivity and generate images, a “monster” has emerged that has changed the rules of the game forever, but this time on a different level, as major financial institutions experienced one of their biggest crises days before and realized that their fear of what was coming had actually come.

In early April this year, Anthropic announced its latest and most powerful model for artificial intelligence that differs from the traditional “Claude” series, which is the “Mythos” model. The aim of its launch was to be a highly capable “cyber agent”, as it is distinguished by its ability to discover security vulnerabilities, which helps companies fill them.

But “Methos” constituted a real terror for major global banks, and triggered a state of extreme alert within the US Federal Reserve and the European Commission, after it demonstrated a terrifying ability to identify security vulnerabilities that had been hidden in banking software codes for three decades.

What we are talking about here is not just an auxiliary tool, but rather a digital entity capable of strategic thinking to penetrate the most complex digital fortresses in a few seconds.

What is the “Mythos” model?

“Methos” is technically classified as a fifth-generation large linguistic model (LLM), developed by the American company “Anthropic” in cooperation with security research bodies.

According to a leaked report published by the American magazine Wired in April, “Methos” is the most advanced version of the “Claude” family of models, but it has been extensively trained in “reverse engineering” and analyzing complex software codes.

Unlike previous models that provided general advice, “Methos” has what is called “independent cyber reasoning,” meaning that it does not need precise human guidance. Rather, it is enough to give it a website address or a copy of a software code, and it will begin searching for hidden “cracks” on its own.

How does Mythos work?

Methos operates through three processing layers that make it superior to any traditional defense system:

1- Deep scanning and pattern recognition (Deep Scanning)
According to a research paper from Stanford University, “Methos” uses a mechanism called “multidimensional attention”, whereby it can read millions of lines of code in fractions of a second, not only to search for known errors, but also to predict how parts of the code will interact in ways that were not intended by the original programmers.

2- Auto-Exploit Generation
Once a vulnerability such as the Zero-Day vulnerability is found, Methos immediately writes the “offensive code.” In a documented experiment conducted by the Defense Advanced Research Projects Agency (DARPA), the model was able to penetrate a fully updated Linux operating system in less than 4 minutes, a process that would have taken the most skilled hacker teams weeks of research.

3- Overcoming behavioral defenses
Banks typically rely on systems that detect strange patterns of attacks, but Methos is smart enough to camouflage its attacks to look like normal traffic, making it “invisible” to traditional cyber radar systems.

What kind of danger does “Mythos” pose?

The terror caused by the “Mythos” in the financial sector lies in three fundamental reasons reported by the British newspaper, the Financial Times, which are:

– The danger of “Legacy Systems”
Major banks, such as the British Bank Barclays or the American JP Morgan, rely on very old software systems written in languages ​​such as COBOL, and these systems are so complex that current programmers are afraid to touch them, but the new model, “Methos”, was able to understand these ancient codes and find security vulnerabilities that had been dormant since the eighties and nineties.

– Simultaneous attack speed
“Methos” can manage thousands of attacks simultaneously, like an army of hackers consisting of ten thousand people working in perfect harmony and at the speed of light, and this type of attack could lead to the collapse of the global banking system in hours if directed at financial clearing points.

– The “time factor” disappears.
Previously, banks had hours or days to respond after detecting a hacking attempt, but with Mythos, hacking, theft and data encryption are happening simultaneously, eliminating any chance of a human response.

International warnings…why now?

Stern warnings were issued by Interpol and the World Economic Forum regarding “METHOS”, and the reason is not only the model’s ability, but also the “democracy of hacking”, as if this model is leaked or sold and any amateur hacker obtains it, he will suddenly turn into a “cyber-terrorist” who possesses the capabilities of a superpower.

The US National Security Agency (NSA) also warned that “Methos” represents an “existential change” in the concept of the digital Cold War, where power is no longer the number of missiles, but rather the number of “parameters” in the artificial intelligence model.

Why was it introduced and what is the purpose?

Some may wonder, if Mythos is so dangerous, why did Anthropic create it? This question is what prompted the company’s technical director to respond at the Black Hat conference this year, which is one of the most famous and important global conferences specialized in information security and cybersecurity, where he said that the goal of raising it is for Anthropic to be “the first to discover vulnerabilities before enemies discover them,” and therefore the declared goal is “proactive defense.”

But banks are currently using “Methos” within a controlled and restricted environment, so that it is the “security auditor.” He is currently attacking the bank and discovering vulnerabilities, and then the programmers close them. This is called a digital vaccination process, where the virus is used to make the vaccine.

The new era of “digital sovereignty”.

Experts do not consider the emergence of “Methos” just a passing technical event, but rather an announcement of the end of the era of “security by ambiguity,” as they put it. It is no longer enough to hide your code to be safe. In the age of artificial intelligence, everything is transparent and readable.

Therefore, observers confirm that the terror that banks are experiencing stems from the fact that “Methos” revealed the weakness of the digital systems that their systems follow, but at the same time it gave them the last opportunity to fortify themselves.

But the fear today will not be of new models, but rather of being too late to realize that the protection rules that banks are accustomed to as protection systems have disappeared, and are being replaced by new systems created by artificial intelligence and with programming complexities, the lack of knowledge of which makes it difficult to control them in the future.