ECB warns banks about cyberattacks using Antrophic's Mythos AI model

The European Central Bank (ECB) is urging banks across the eurozone to urgently prepare for potential cyberattacks using AI tools such as Anthropic’s Mythos or similar systems. The warning was issued by ECB Executive Board member Frank Elderson, a Dutch attorney, in a central bank newsletter on Wednesday. At the same time, French artificial intelligence start-up Mistral, which has support from Dutch tech giant ASML, is in talks with European banks about deploying its alternative to Mythos, news agency Bloomberg reported.

Mythos can quickly detect security flaws, which could help banks strengthen their own systems. However, cybersecurity experts warn that if the same technology is used by criminals, it could also be turned against financial institutions, making bank systems more vulnerable to attack.

A further challenge for European banks is that access to Mythos remains highly restricted, preventing them from using the tool directly. Elderson says this should not be an excuse for inaction. “Lack of access is no excuse for doing nothing. On the contrary, it makes it even more important for banks to act now,” he said.

He added that security flaws once seen as low priority, and typically patched later, should now be treated as urgent and resolved immediately. Mythos is capable of “rapidly combining seemingly minor vulnerabilities into serious attacks that previously could only be carried out by teams of experts working for several days,” Elderson pointed out.

“This is an urgent situation; we are not dealing with a long-term scenario. It is uncertain how long it will take before these capabilities become more widely available.” After Mythos, there will likely be new tools that enable even more aggressive cyberattacks, he predicted.

Mistral was already working with banks on the use of AI to detect security vulnerabilities, similar to what Mythos was designed to do. Sources told Bloomberg that the French company now plans to launch a standard version of that technology to enable broad sales.

Anthropic currently makes Mythos available only to a select group of partners, including tech companies, cybersecurity firms, and banks. They can test the capabilities of the AI ​​model and strengthen their own security based on the results.

Mistral declined to comment substantively on questions from Bloomberg. Earlier this week, Mistral CEO Arthur Mensch told French parliamentarians how important he considers technological autonomy in Europe.

"It is impossible to have the source code of the French army checked by Mythos. That creates such an irreversible dependency that we absolutely must find solutions," he said. He said that Mistral and other American and Chinese tech companies are already capable of detecting security vulnerabilities and spoke of "scaremongering" surrounding Mythos.

Mistral counts the British bank HSBC and the French BNP Paribas among its clients. The Dutch chip machine manufacturer ASML is a major investor in Mistral.