Malware uses language models to estimate the value of victims

Security experts have documented a new incident involving the use of generative artificial intelligence in cybercrime. The VOLTA-AUTODOXER malware uses a language model to rank compromised users based on their financial value.

The infostealer collects data from compromised devices and then sends the information to an algorithm for analysis. The automated assessment system processes the victim's purchase history, workplace, job title, and other personal details. Based on these parameters, the malware determines the priority for further attacks on specific accounts.

This approach allows attackers to automate the reconnaissance phase and focus on the wealthiest or corporate targets. The practice of using AI tools in malware is changing the approach to cyberattacks, making them more targeted and resource-efficient for the organizers.