Research · curated 12 Aug 2026
A systematic literature review of large language models in phishing attack generation and detection
First reported sciencedirect.com
Coverage timeline
Single-source research — first reported, latest, and curated coincide.
Why it matters
The review documents how LLMs are accelerating and automating phishing content generation, including multimodal attacks, which defenders must anticipate as GenAI tools lower the skill barrier for adversaries.
A systematic literature review (PRISMA methodology, 36 studies from 2023-2025) examines the dual role of large language models in both generating and detecting phishing attacks. The review finds LLMs, especially GPT-based models, lower the technical barrier for attackers by producing coherent, persuasive phishing email and website content, while also strengthening detection through analysis of textual and visual content, often outperforming traditional machine-learning approaches.