← All news
The Hacker News

Stealthy Linux Cryptominer 'SANSAR' Hides GPU Load via Modified NVIDIA Library

Stealthy Linux Cryptominer 'SANSAR' Hides GPU Load via Modified NVIDIA Library

Security researchers have identified a financially motivated threat activity cluster, tracked as REF3364, running a Linux-focused cryptocurrency mining operation. The campaign's defining feature is its use of a modified NVIDIA library component that falsifies the readings system monitoring tools rely on to detect abnormally high GPU workloads, allowing the malware—dubbed SANSAR—to mine cryptocurrency while masking the telltale signs of heavy resource consumption.

Cryptomining malware remains one of the most common forms of commodity cybercrime because it offers attackers a low-risk, recurring revenue stream: once a system is compromised, it can be silently monetized for as long as it goes undetected, without the operational overhead of data exfiltration or ransom negotiation. GPU-based miners are particularly attractive to attackers because graphics cards can process certain mining algorithms far more efficiently than CPUs, making compromised servers with discrete GPUs—common in cloud, gaming, and AI/ML infrastructure—high-value targets. The core challenge for defenders has long been that sustained high GPU or CPU utilization is one of the easiest indicators of compromise to spot, which is why threat actors increasingly invest in techniques to spoof or suppress those metrics rather than simply hiding the binary itself.

By manipulating a legitimate vendor library rather than relying solely on a custom payload, this approach raises the bar for detection, since monitoring tools and administrators that trust standard NVIDIA utilities for hardware telemetry may be fed falsified data. This reflects a broader trend of malware authors targeting the trust placed in system and vendor tooling, rather than just the operating system or user-space processes, to blind defenders during live compromise.

Organizations running Linux systems with GPU hardware, particularly in cloud and virtualized environments, should verify the integrity of GPU driver and monitoring libraries against known-good hashes, audit for unauthorized modifications to system libraries, and supplement vendor-reported telemetry with independent, out-of-band resource monitoring (such as hypervisor-level or network-based utilization checks) that cannot be tampered with from within the compromised host. Unexplained billing spikes in cloud environments, unexpected outbound connections to mining pools, and discrepancies between reported and actual power/thermal consumption can serve as additional detection signals when GPU utilization metrics themselves are untrustworthy.

#malware#cryptomining#linux#gpu security

Source: The Hacker News

Read the original report ↗

This is an original summary written by CyberSkii for reference. All rights to the underlying reporting remain with the publisher linked above.

Need help? Contact us