PortSwigger: The Evolution of LLMs into Autonomous Attack Inventors
Research from PortSwigger, led by James Kettle, demonstrates a paradigm shift in Large Language Model (LLM) utilization within the cybersecurity domain. Moving beyond simple code completion, LLMs are being leveraged as autonomous security researchers capable of discovering novel, zero-day attack vectors. By employing intelligent permutation of attack patterns and high-volume hypothesis testing, these models can generate complex, non-obvious payloads, such as advanced HTTP Request Smuggling variants. This transition from manual payload crafting to the orchestration of autonomous agents significantly reduces the time-to-discovery for sophisticated logic flaws and lowers the technical barrier for executing multi-stage, complex attack chains.
PentestGPT
PentestGPT is an open-source agentic framework designed to automate the end-to-end penetration testing lifecycle. Unlike traditional LLM-based assistants that function as passive consultants, PentestGPT utilizes a modular three-tier architecture—Reasoning, Execution, and Planning/Knowledge—to maintain state and logical continuity across multi-step attack chains. The framework integrates with toolsets like Claude Code and standard security utilities through an orchestration layer, enabling autonomous reconnaissance, vulnerability discovery, and exploit execution. Benchmarks demonstrate a 228.6% improvement in task completion efficiency over standalone GPT-3.5, significantly reducing the necessity for human-in-the-loop intervention during complex security engagements.