Researchers say credential-stealing campaign used AI to build evasion ‘at every stage’
Researchers say credential-stealing campaign used AI to build evasion ‘at every stage’
https://cyberscoop.com/deepload-ai-malware-obfuscation-at-every-stage-reliaquest/
Publish Date: 2026-03-30 14:28:17
Source Domain: cyberscoop.com
Summary
A new malware campaign, dubbed “DeepLoad,” is exploiting advanced techniques, including artificial intelligence, to infiltrate and persist within enterprise IT environments, stealing credentials through various means including real-time keylogging. ReliaQuest AI’s analysis, by Thassanai McCabe and Andrew Currie, highlights the malware’s sophisticated obfuscation methods that can evade traditional security measures. The attackers employ social engineering techniques to deliver the malware, which burrows deeply into infected systems through heavy code padding and operating behind overlooked Windows processes. As the attackers show the capability to continually update their methods, experts warn that AI-driven obfuscation makes it hard for organizations to adapt their current static security practices. McCabe and Currie emphasize the need to shift focus towards behavioral, runtime detection to combat such evolving threats, suggesting that file-based scanning is becoming increasingly less effective against AI-enhanced cyberattacks.
Key Points:
- DeepLoad uses AI to evade traditional security measures, turning user actions into persistent credential-stealing access.
- The malware spreads through social engineering and employs sophisticated obfuscation techniques that likely involve AI.
- DeepLoad persists even after security efforts to block it, spreading to connected USB drives and re-executing three days later even after cleanup.
- Traditional static cybersecurity practices like signature-based detection may become obsolete as AI continues to create unique attack variations.
- Experts recommend shifting focus to behavioral and runtime detection methods over file-based scanning to counteract these advanced threats.
