A new fear has grown in European financial institutions in recent weeks. Along with the instability caused by the geopolitical context, new technological models are posing a practically unexpected challenge for banking supervisors. For this reason, at their meeting this Monday, the Eurogroup decided to dedicate an item on the agenda to the implications of the Mythos Artificial Intelligence system from the American company Anthropic.
Spain asks for coordination
Upon his arrival at the Eurogroup meeting, the vice president and minister of Economy, Carlos Cuerpo, expressed the need for the continent to address the threats posed by new artificial intelligence models in terms of their ability to find flaws or backdoors in practically “all institutions”. "The institutions in the United States that are seeing how this model affects them are not only limited to the financial sector, but are part of the economy as a whole," he explained.
The number two of the Government has confessed that the evolution of these disruptive models is occurring at a much faster pace than could have been imagined just a few months ago, so he does not rule out new appearances of systems similar in sophistication. Faced with this scenario, the Spanish recipe involves a key concept: coordination.
What Europeans are looking for is how, with the existing tools on the table, the Twenty-Seven can be able to articulate a strategy by which companies can protect themselves while maintaining access to these models. After the proposal by the chief supervisor of the German Bundesbank, Michael Theurer, in an interview with Reuters, that European banks should have access to Anthropic's latest developments —including Mythos— in order to protect themselves against possible cyberattacks, Cuerpo has argued that “Member States cannot go each one on their own in a uncoordinated manner to try to access this software”.
“We cannot be, in this case, a second-class region when it comes to waiting for our companies to be able to defend themselves from these models and for us to have early access,” the minister has stated in Brussels, with the objective that this access allows the discovery of the existing vulnerabilities in the current systems.
The first European response
The European Central Bank has already made its move and has asked the eurozone partners to begin analyzing the risks associated with these new artificial intelligence models. The concern focuses mainly on key issues for cybersecurity, due to the capacity this type of technology has to penetrate vulnerabilities in computer systems, which can have critical consequences for banking.
From Frankfurt, the institution would be requesting from the entities a reinforcement of their preparation, which could be summarized in the elaboration of new contingency plans and in a more exhaustive evaluation of their vulnerabilities before possible uncertain scenarios. More than an extraordinary response, the body presided over by Christine Lagarde is gathering information through its usual dialogue with the supervised entities.
The objective is to guarantee to what extent banks have effective mechanisms to respond to threats related to the negative use of new tools generated by artificial intelligence. In other words, Brussels and Frankfurt seek to anticipate a scenario in which technology ceases to be solely an ally to also become a vector of systemic risk.
How does this model work?
Anthropic's new product is designed to go one step further in digital security through unprecedented artificial intelligence systems. This has sparked a wave of expectation in the sector, accompanied by growing concern about the possible repercussions in terms of security and international politics.
What has really set off all alarms is its ability to expose vulnerabilities in software, including those known as "zero-day" flaws. These are errors unknown until now that can be exploited before a solution exists, which makes Mythos an extremely powerful tool within the cybersecurity field.
This level of autonomy and analytical capacity would have led the company itself to keep the system out of public access, precisely because of the potential risks associated with its misuse. The sophistication of the model not only allows for the detection of weaknesses, but also for the simulation of complex scenarios in which such vulnerabilities could be exploited in a coordinated manner.
All of this has caused that, in the United States, the system has been classified by defense services as a possible risk to the technological supply chain, given its capacity to identify weak points in critical infrastructures. For his part, the CEO of Anthropic, Dario Amodei, has defended the company's approach, stressing that these technologies must be developed under strict security controls and with clear limits of use.
In this context, Europe faces a strategic dilemma: how to harness the potential of these tools without compromising the stability of its financial system. The answer, as several capitals agree, will inevitably involve strengthening cooperation between states, promoting joint supervision, and accelerating regulatory adaptation to a technological environment that is evolving at an unprecedented speed.