One Step Ahead in Cyber Hide-and-Seek: Automating Malicious Infrastructure Discovery With Graph Neural Networks

One Step Ahead in Cyber Hide-and-Seek: Automating Malicious Infrastructure Discovery With Graph Neural Networks
This article discusses the proactive detection of cyber threats through automated pivoting on known indicators, showcasing three case studies involving phishing campaigns. It highlights the use of a graph neural network (GNN) to uncover new malicious domains and emphasizes the importance of continuous monitoring of threat actors’ evolving indicators. Affected: phishing campaigns, web skimmer campaigns, financial services phishing campaigns

Keypoints :

  • Threat actors leave traces of information when launching large-scale attacks.
  • Automated pivoting can help defenders uncover new attack infrastructure.
  • Three case studies illustrate the effectiveness of this approach: postal services phishing, credit card skimmer campaign, and financial services phishing.
  • Palo Alto Networks provides advanced security measures to protect against these threats.
  • Continuous monitoring of threat actors’ indicators can lead to proactive defense strategies.

MITRE Techniques :

  • T1071.001 – Application Layer Protocol: Threat actors use various domains to communicate with command-and-control (C2) servers.
  • T1070.001 – Indicator Removal on Host: Threat actors may rotate domains to evade detection.
  • T1071.003 – Application Layer Protocol: Use of malware delivery URLs to distribute malicious binaries.
  • T1046 – Network Service Discovery: Identifying co-hosted domains to map out infrastructure.
  • T1583.001 – Acquire Infrastructure: Threat actors register numerous domains to support phishing campaigns.

Indicator of Compromise :

  • [domain] advanced-ip-sccanner[.]com
  • [domain] myipscanner[.]com
  • [domain] myscannappo[.]com
  • [domain] correosespana[.]top
  • [domain] apple.com-ticket[.]info
  • Check the article for all found IoCs.



Full Research: https://unit42.paloaltonetworks.com/graph-neural-networks/