Researcher(s)
- Juan Pablo Chavez Galicia, Computer Science, University of Delaware
Faculty Mentor(s)
- Sunita Chandrasekaran, Data Science Institute, University of Delaware
Abstract
In today’s era of escalating cyberthreats, traditional website security tools often fail to detect sophisticated vulnerabilities due to their rigid scanning methods and inability to analyze long-term security trends. This research demonstrates how integrating artificial intelligence (AI) with conventional security scanning can significantly improve vulnerability detection, particularly for security header misconfigurations, SSL/TLS weaknesses, and gradual security degradation. By using large language models (LLMs) like Ollama’s DeepSeek, the system provides intelligent analysis of security risks, evolving threats, and prioritized remediation steps, going far beyond basic vulnerability detection.
The AI-enhanced approach delivers multiple key advantages over traditional scanners:
- Historical Security Analysis: The system tracks and interprets changes in security headers over time, identifying dangerous trends that single-point scans miss. This reveals slowly developing vulnerabilities that attackers could exploit.
- Reduced False Positives: By understanding context, the AI distinguishes between intentional security configurations and actual vulnerabilities, significantly cutting down on misleading alerts.
- Actionable Reporting: Clear, prioritized recommendations and plain-language explanations make findings accessible to both security teams and non-technical stakeholders.
- Continuous Learning: The system improves its detection capabilities over time as it processes more security data.
Looking ahead, this framework could evolve to support real-time security monitoring, automated remediation suggestions, and integration with DevOps pipelines for continuous protection. As cyber threats grow more sophisticated, AI-augmented security scanning represents not just an improvement, but a necessary evolution in website protection – offering smarter, more adaptive defense against modern web threats.