Large amounts of stolen data from bank customers can be found on relevant platforms on the Dark Web. There are so many that IT security experts like Benjamin Mejri from Evolution Security assume that hackers captured them using artificial intelligence.
“Many actors are using AI. The attacks are becoming more professional, they are faster and a higher number of targets are being attacked,” said the expert.
And cyber gangsters are evaluating bank customer data from older data leaks and phishing attacks using AI. They also attack banks directly.
Mejri observes that attackers are increasingly using AI in a targeted manner: "Actors are increasingly using specifically trained, unregulated Large Language Models (LLMs) to further develop existing attack campaigns, optimize attack vectors and improve the analysis and evaluation of captured data in the context of digital attacks."
Disturbing case in South Korea
A case from South Korea is worrying. As the Korean news channel SBS News reports, seven financial institutions have fallen victim to a suspected AI attack. On October 4th there was even an emergency meeting of the local financial regulator. IT security expert Adam Meyers from the company CrowdStrike investigated the case and explained in an interview with Plusminus that the attack was started from an IP address in China.
According to Meyers, an attacker without any special IT knowledge commissioned an unregulated Chinese AI to search for vulnerabilities in banks in South Korea. "Once the AI had access, it moved very quickly through the IT environment, tracking down data and passing it on to the actor behind it." According to Meyers, the attacker also asked the AI to give him tips on how to market the stolen data.
How AI attacks on banks work
Oliver Keizers from the IT security company Horizon3 describes how AI attacks work. If AI systems were to find a vulnerability, additional tools could independently expand the attack: "They are fully automated attacks and as soon as the first open door is found somewhere, the 'special teams' go in and expand the whole thing."
The attacker AI should then build so-called exploits for identified vulnerabilities, attacker tools for the respective vulnerabilities. For Oliver Keizers, who uses AI on behalf of banks to find vulnerabilities, it's about being faster than the hackers.
Are Germany's banks well protected against AI attacks?
The decisive factor for security is who finds the security gaps first. Claudia Buch, head of the ECB's banking supervision department, called on the institutions in the summer to "immediately assess the effects of the changed threat situation and present a comprehensive action plan with concrete measures."
When asked, the German banking industry explains that the action plans are not public.
100 million euros in damage after cyber attack
According to its own information, the French bank Crédit Mutuel Arkéa suffered damage of around 100 million euros as a result of a hacker attack. The investigation is ongoing "in order to precisely determine the methods used by the attackers," explains a spokeswoman for Crédit Mutuel Arkéa when asked. The bank cannot confirm the use of AI with certainty.
IT security experts assume that cybercriminals will no longer use AI to attack banks and their customers.
