Hacker Utilizes AI-Based Malware for Bug Bounty Exploits

Researchers at CrowdStrike have uncovered a hacker posing as a bug bounty hunter who exploited AI-generated malware to gain access to systems and identify vulnerabilities.

The individual, known by the alias PhantomRaven, disguised the malicious code within legitimate libraries found in npm repositories. Once installed, these packages activated a stealer on the device.

Malicious npm libraries. Source: CrowdStrike.

The hacker collected system data and other information to access corporate resources. CrowdStrike described the tool as relatively simple, and the distribution method as a classic software supply chain attack.

According to CrowdStrike, the stealer's code was likely generated using a large language model (LLM), as indicated by verbose comments, template-like snippets, and statistical token patterns within the script.

The company speculated that a financially motivated hunter was behind these attacks, using the access gained to infiltrate corporate systems, search for vulnerabilities, and submit reports to bug bounty programs.

PhantomRaven had been active in the network since 2022, with mentions of rewards from at least nine companies across technology, retail, and hospitality sectors on his X page. He also collaborated with platforms for white hat hackers, including Bugcrowd, Intigriti, YesWeHack, HackenProof, and HackerOne.

In November 2025, the hacker personally contacted a potential victim, claiming to have found a compromise on their device, attributing it to another campaign that employed malicious npm packages.

Additionally, the hacker attempted to broaden his attack vector by uploading code to PyPI but faced rejection. An analysis of the remaining Python files revealed similar stealer code.

Researchers found no evidence of the stolen data being sold on dark web marketplaces, supporting the theory that access was solely used for vulnerability discovery and bug bounty earnings.

The CrowdStrike team noted that they had mitigated several incidents involving PhantomRaven.

Experts stated that the attack method ceased to function in newer versions of npm, as the package manager has changed its behavior to block installation scripts by default until developers manually permit their execution.

It is worth noting that in September 2026, F6 analysts highlighted the rapid adoption of AI among hackers, with AI-based tools becoming integrated into real cyberattacks—from initial reconnaissance to the generation of malicious code.