AI-Generated Adversarial Scripts for Malware Detector Validation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing cybersecurity systems face challenges in effectively testing the robustness of AI-driven malware detectors against adversarial attacks due to the complexity and diversity of malware detectors and the rapidly evolving nature of adversarial techniques, leading to a lack of sufficient validated adversarial scripts for comprehensive testing.

Innovation Solution

An AI model is used to automatically generate adversarial scripts that obfuscate malicious behaviors, which are then tested against malware detectors to determine their robustness, with validation through dynamic and static analysis, and feedback loops to refine the scripts based on detector responses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual creation of adversarial scripts is used, then validation precision can be maintained, but productivity and quantity of tested scripts decrease

Engineering Contradiction:
Improvevalidation precisionVSAvoidproductivity
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system uses AI models to automatically generate adversarial scripts that test malware detectors, allowing the system to self-test and self-improve without requiring manual creation of each test script

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system incorporates feedback loops where the results of adversarial script testing are used to refine and improve future script generation, creating a continuous improvement cycle that maintains validation precision while increasing productivity

Inventive Principle:
Principle #23Feedback

2Reliability

If comprehensive testing of malware detectors is performed, then reliability of security systems improves, but device complexity and time requirements increase

Engineering Contradiction:
ImprovereliabilityVSAvoiddevice complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The testing process is segmented into multiple stages including script generation, validation through dynamic and static analysis, and feedback refinement, allowing comprehensive testing to be broken down into manageable components that reduce overall system complexity

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Adversarial scripts are pre-generated and validated before being used to test malware detectors, allowing the system to prepare comprehensive test cases in advance and reducing the complexity of the testing process during execution

Inventive Principle:
Principle #10Preliminary action

3Adaptability or versatility

If adversarial techniques are rapidly evolved, then security against new threats improves, but difficulty of detecting and measuring increases

Engineering Contradiction:
ImproveadaptabilityVSAvoiddifficulty of detecting and measuring
Core Design Contradiction:
Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The system dynamically generates and validates adversarial scripts using AI models that can adapt to evolving malware detection techniques, allowing the testing system to keep pace with rapidly changing security landscapes

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system replaces manual analysis methods with automated AI-based generation and validation processes, making it easier to detect and measure the effectiveness of adversarial techniques while maintaining adaptability to new threats

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4625259A1Adversarial script generation using artificial intelligence models
Publication Date: 2025.10.01 CROWDSTRIKE
  • EP4625259A1 patent drawingFigure 1
  • EP4625259A1 patent drawingFigure 2A~2B
  • EP4625259A1 patent drawingFigure 3

AI summary

The present disclosure provides an approach of providing, to an artificial intelligence (AI) model, a malicious script that includes a malicious behavior. The AI model is configured to modify software code of the malicious script to produce modified software code that obfuscates the malicious behavior. The approach produces, by a processing device using the AI model, an adversarial script that includes the modified software code that obfuscates the malicious behavior. In turn, the approach initiates a malware detector to test the adversarial script.