ShadowRay 2.0 Exploits Unpatched Ray Flaw to Build Self-Spreading GPU Cryptomining Botnet

ShadowRay 2.0 is described as an evolution of a prior wave that targeted unpatched Ray installations to turn infected GPU clusters into a self-spreading cryptomining botnet. According to the report, the campaign exploits CVE-2023-48022, a critical missing authentication bug in the Ray AI framework, with a CVSS score of 9.8, to hijack susceptible instances for illicit mining with XMRig.

The flaw has remained unpatched due to a long-standing design decision that Ray runs in isolated networks and binds to trusted code, the article notes. Attacks involve submitting malicious jobs via an unauthenticated Ray Job Submission API (the “/api/jobs/” endpoint) on exposed dashboards, enabling a worm-like propagation to other Ray dashboards.

Researchers Avi Lumelsky and Gal Elbaz describe the threat as turning Ray’s orchestration features into a self-propagating global cryptojacking operation, spreading across exposed clusters and enabling persistence through cron jobs that fetch updates from GitLab. The campaign has reportedly used GitLab and GitHub to host payloads, with more than 230,500 Ray servers publicly accessible at risk. Anyscale has released a Ray Open Ports Checker to help verify cluster configurations and mitigate exposure.

The attackers’ activity has included attempting DoS-style tactics and disguising processes to resemble legitimate Linux kernel worker services, while the data suggests the operation may have been active since September 2024.