Cryptojacking Campaign Targets PostgreSQL Servers Via Unauthorized Access

Article Highlights
Off On

A concerning ongoing campaign has been identified, targeting exposed PostgreSQL instances to deploy cryptocurrency miners through unauthorized access. This recent cyber intrusion has been closely monitored by security firm Wiz, which has labeled this malicious activity as a variant of the intrusion initially discovered by Aqua Security last year. Using malware named PG_MEM, the campaign is attributed to the threat actor JINX-0126. This operation showcases sophisticated defense evasion techniques, such as deploying binaries with unique hashes for each target and executing miner payloads filelessly to evade detection.

Unauthorized Access and Exploits

The campaign has successfully compromised over 1,500 PostgreSQL servers, primarily taking advantage of systems with weak or easily guessable credentials. One of the key tactics employed by the attackers involves abusing the COPY … FROM PROGRAM SQL command to execute shell commands, allowing them to gain an initial foothold on the targeted server. They further use a Base64-encoded shell script to eliminate competing cryptocurrency miners and install a malicious binary named PG_CORE. This method ensures continued control over the infected system, enabling the attackers to maintain their cryptojacking activities without interruption.

An obfuscated Golang binary, deceitfully named postmaster to mimic the legitimate PostgreSQL server, aids persistence. This binary establishes itself via cron jobs, creates a privileged role within the database, and writes another binary, cpu_hu, to disk. The binary cpu_hu, in turn, downloads the latest version of the XMRig miner from GitHub and launches it filelessly using the memfd technique, a tactic known for its efficiency in avoiding detection. Each compromised machine becomes uniquely assigned as a mining worker, contributing to the attacker’s overall mining operations.

Indicators and Impacts of the Campaign

Evidence indicates that the threat actor assigns a unique identifier to each mining worker, with three wallets showing links to approximately 550 workers each. This alignment implies that the campaign may involve more than 1,500 compromised machines. The wide-reaching impact highlights the significant risk posed by publicly exposed PostgreSQL instances, which often become opportunistic targets for such malicious campaigns due to their vulnerabilities.

The campaign’s distinctive characteristics include fileless payload execution, unique binary hashes per target, and sophisticated persistence mechanisms. This level of intricacy reflects the evolving strategies employed by JINX-0126, who have demonstrated a refined approach to cryptojacking. These developments underscore the critical need for robust security practices and proactive measures to safeguard cloud environments and database servers from exploitation.

Strategic Takeaways

A significant ongoing cyber campaign has been detected, targeting unprotected PostgreSQL databases to install cryptocurrency mining software through unauthorized access. Security firm Wiz has been closely monitoring this recent cyber intrusion, identifying it as a variant of an attack first uncovered by Aqua Security last year. The malware, named PG_MEM, is linked to the threat actor known as JINX-0126. This sophisticated operation uses advanced techniques to avoid detection, such as deploying binaries with unique hashes for each target and executing their miner payloads without creating files, rendering traditional security measures less effective. The attackers’ ability to adapt and evolve their strategies highlights the critical need for robust security measures to protect database instances from such threats. Organizations are urged to secure their PostgreSQL instances by applying necessary security patches, monitoring unusual activities, and implementing best practices to avoid falling victim to these sophisticated attacks.

Explore more

NHS Federated Data Platform – Review

While the global financial landscape reacts with fervor to the immense valuation of enterprise reasoning software, the National Health Service currently navigates a paradoxical reality where it owns one of the world’s most advanced data engines yet struggles to activate its full operational power across its vast network of trusts. The NHS Federated Data Platform (FDP) is not merely a

Can Apple Protect Mac Privacy From Autonomous AI Agents?

The seamless transition of artificial intelligence from a passive search tool to an autonomous operator marks a pivotal shift in how individuals interact with their personal computers. This evolution promises a future where digital assistants manage complex workflows, yet it simultaneously erodes the traditional barriers that once kept sensitive user data behind locked gates. As of 2026, the arrival of

Citrix Patches Actively Exploited NetScaler Zero-Day

Modern corporate networks depend so heavily on seamless authentication that even a brief interruption in Gateway services can freeze global operations and leave remote workforces stranded without access. Security leaders are now confronting a significant challenge involving memory mismanagement in primary entry points that requires immediate attention to maintain connectivity. Overview of the NetScaler Zero-Day Vulnerability CVE-2026-88779 is a high-severity

How Is AI-Generated Code Changing Linux 7.3 Development?

The massive complexity of the Linux kernel now exceeds 40 million lines of code, a scale that has fundamentally altered the way developers interact with one of the most critical pieces of digital infrastructure in existence today. This sprawling codebase represents a culmination of decades of collective human effort, yet the 7.3 development cycle signals a distinct departure from traditional

pgEdge Launches Starfleet to Bridge the AI Production Gap

The current enterprise technology landscape is defined by a frantic and often disorganized race to move from experimental concepts toward functional, value-driven applications that can survive the rigors of a global market. While developers are successfully building sophisticated generative AI and agentic workflows in isolated environments, they frequently encounter a significant wall when attempting to deploy these tools at a