Analysis
Large Language Models (LLMs) and Generative AI in Cybersecurity and Privacy: A Survey of Dual-Use Risks, AI-Generated Malware, Explainability, and Defensive Strategies
Publication date not yet evaluated · Added · 4 reports arxiv.org
Page published
Coverage timeline
Why it matters
The survey synthesizes the evolving dual-use landscape where LLMs power both automated defense and increasingly automated attacks, giving defenders a consolidated reference on AI-driven threats and countermeasures.
A survey paper by Kiarash Ahi and Saeed Valizadeh reviews over 70 academic papers, industry reports, and technical documents on the dual-use of LLMs and generative AI in cybersecurity, covering AI-generated malware, zero-day detection, explainable AI, and defensive strategies. Drawing on case studies from platforms like Google Play Protect, Microsoft Defender, and Hugging Face, it offers recommendations including model watermarking, adversarial defense, and cross-industry collaboration.