Large Language Models (LLMs) and Generative AI in Cybersecurity and Privacy: A Survey of Dual-Use Risks, AI-Generated Malware, Explainability, and Defensive Strategies

2026-07-08 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Expert, quick

Summary

A comprehensive survey published on 2026-07-08 examines the dual-use risks and defensive strategies associated with Large Language Models (LLMs) and generative AI (GenAI) in cybersecurity and privacy. This analysis covers systems like ChatGPT, Claude, Gemini, LLaMA, Copilot, and Stable Diffusion from vendors including OpenAI, Anthropic, Google, Meta, Microsoft, and Stability AI. The survey highlights how these technologies enable both automated defense, such as real-time threat detection and secure code generation, and sophisticated attacks, noting that LLM-generated malware is projected to account for 50% of detected threats by 2025, a significant increase from 2% in 2021. Drawing on over 70 academic papers and industry reports, the work synthesizes insights from platforms like Google Play Protect, Microsoft Defender, AWS, and GitHub, addressing topics such as zero-day detection, DevSecOps, federated learning, synthetic content analysis, and explainable AI (XAI).

Key takeaway

For security leaders and engineers navigating the complex challenges of AI-driven cybersecurity, you must prioritize next-generation security frameworks. Your teams should implement model watermarking and adversarial defense strategies to counter the rapid surge in LLM-generated malware, which is projected to reach 50% of detected threats by 2025. Proactive cross-industry collaboration is essential to establish secure, scalable LLM systems and ensure trustworthy AI deployment.

Key insights

LLMs present significant dual-use challenges in cybersecurity, necessitating advanced defensive frameworks and responsible deployment.

Principles

• LLMs automate both defense and attack.
• AI-generated malware is rapidly increasing.
• Responsible LLM deployment needs transparency.

Method

The survey synthesizes insights from over 70 academic papers and industry reports, reviewing real-world case studies across major platforms to identify beneficial and malicious LLM applications.

In practice

• Implement model watermarking.
• Develop adversarial defense strategies.
• Engage in cross-industry collaboration.

Topics

• Large Language Models
• Generative AI
• Cybersecurity Risks
• AI-Generated Malware
• Explainable AI
• Defensive Strategies

Best for: CTO, VP of Engineering/Data, Executive, AI Scientist, AI Security Engineer, Director of AI/ML

Related on AIssential

• AI-Generated PowerShell Malware: An Experimental Framework and Dataset — LLMs can generate PowerShell malware highly similar to real-world samples, posing a significant cybersecurity threat.
• SLMs, LLMs, and the Real Difference That Matters in DSPM — The real DSPM difference is not model size, but whether models predict fixed patterns or reason generatively.
• The Emergence of Autonomous Penetration Capabilities in Large Language Model-Powered AI Systems — LLM-powered AI systems exhibit autonomous penetration capabilities, with success rates up to 69.3% in realistic evaluations, improving with model advancement.
• MLWhiz Weekly Recsys/ML/GenAI Newsletter # 6 — AI capabilities are rapidly outrunning existing institutional frameworks in cybersecurity, coding, and language design.
• Can Small GenAI Language Models Rival Large Language Models in Understanding Application Behavior? — Small GenAI models can rival larger LLMs in malware detection, offering efficiency for resource-constrained environments.
• AI Worm — The AI worm prototype integrates an LLM for autonomous, decentralized propagation and multi-attack generation, posing a new cyber threat.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.