Hacker under the pseudonym PhantomRaven has been using an AI stealer to find vulnerabilities in systems since 2022. He developed and distributed malicious software disguised as legitimate libraries in npm repositories. After installing such packages, the malicious code would run on devices, collecting system data and other information to gain access to corporate resources. Researchers at CrowdStrike believe that the stealer's code was most likely generated using LLM, as indicated by verbose comments, template fragments, and token patterns in the script. The PhantomRaven attack represents a classic software supply chain attack scheme, confirmed by code analysis and its structure.
PhantomRaven actively collaborated with various platforms for white hat hackers, including Bugcrowd, Intigriti, and HackerOne, and received rewards from at least nine companies operating in the technology, retail, and hospitality sectors. In November 2025, he even reached out directly to one of his potential victims, claiming to have discovered a device compromise, while attempting to link it to another campaign using malicious npm packages. Despite his activity, researchers found no evidence of selling stolen data on shadow markets, indicating that the accesses were used solely for finding vulnerabilities and obtaining rewards.
CrowdStrike reported that the attack used by PhantomRaven stopped working in new versions of npm, as changes in the behavior of the package manager block the execution of installation scripts by default without the developer's permission. In September 2026, F6 analysts also noted the rapid spread of AI tools among hackers, raising the question of ensuring reliable protection against such threats.
Related articles
