Germany on Alert Over Anthropic's AI Mythos

Germany Alarmed by Claude Mythos

Germany has recently begun to closely monitor Claude Mythos, an artificial intelligence model developed by Anthropic. This model is particularly effective at detecting security vulnerabilities within software, a strength that could turn into a threat if this technology fell into the wrong hands. According to a Reuters report published this week, German banks and relevant authorities are currently assessing the risks associated with this model. Kolja Gabriel, a board member of the German Banking Association, stated that the group is actively consulting with cybersecurity experts from banks, as well as the Ministry of Finance, to anticipate potential threats.

The use of Mythos is currently controlled by companies specializing in cybersecurity, which utilize it to quickly identify and rectify potential vulnerabilities. Gabriel also mentioned that several software updates are planned in the near future, and the situation is being closely monitored. Meanwhile, BaFin, the financial sector supervisory authority in Germany, has warned that financial institutions must prepare for the possibility of new security vulnerabilities that will require rapid fixes.

Although Claude Mythos was officially announced in April, Anthropic has not yet launched this model. Instead, the company has initiated the Glasswing project, which aims to secure critical software by giving an edge to those who protect it. This project involves collaborations with tech giants such as Amazon, Apple, Google, Microsoft, Nvidia, and JP Morgan Chase. Even before the announcement of this project, Claude Mythos had already identified thousands of severe security vulnerabilities, including in widely used operating systems and browsers.

Europe and France Confronting Mythos

At the European Union level, the ECB has already taken action. According to Reuters, it is currently questioning European banks to assess their level of preparedness against this new threat. The European Commission has welcomed Anthropic's decision not to immediately deploy the Mythos AI model.

A representative from Brussels, quoted by Politico, stated: "Given the large-scale cyber risk posed by this model, we applaud this gradual deployment." However, in France, no official statement has yet been made. Nevertheless, given the magnitude of the risks and the media attention surrounding Claude Mythos, it is likely that discussions are taking place behind the scenes. Anthropic has also warned that similar models could soon be developed by other players. "With the current pace of AI advancements, these capabilities could quickly spread, potentially beyond companies committed to using them securely. The consequences for economies, public safety, and national security could be severe," the company stated.

A Race Against Time

Anthropic's statements suggest a genuine race against time. Indeed, according to the company, a model with capabilities similar to those of Claude Mythos could soon be accessible to malicious actors. The goal is therefore to prepare critical software for this eventuality.

In Germany, banks are already taking steps to assess the risks associated with Claude Mythos, the AI from Anthropic recognized for its ability to identify security vulnerabilities. The ECB has also begun questioning European banks about their level of preparedness for this potential threat. Although Mythos is an extremely powerful AI, Anthropic has not yet decided to launch it, aware that its level of competence could lead to a wave of cyberattacks if misused.