Up until now, artificial intelligence agents — AI systems that can work independently toward specific goals without constant human guidance — demonstrated strong capabilities in software development and web navigation, but their effectiveness in cybersecurity remained suspect.

That may soon change.

That is because an AI agent capable of autonomously solving complex cybersecurity challenges is now in development at NYU Tandon School of Engineering, NYU Abu Dhabi and other universities and the researchers call it EnIGMA.

“EnIGMA is about using Large Language Model (LLM) agents for cybersecurity applications,” said Meet Udeshi, a NYU Tandon Ph.D. student and co-author of the research.

To build EnIGMA, the researchers started with an existing framework called SWE-agent, originally designed for software engineering tasks. However, cybersecurity challenges required specialized tools that didn’t exist in previous AI systems.

Schneider Bold
Restructured Interfaces

“We have to restructure those interfaces to feed it into an LLM properly. So we’ve done that for a couple of cybersecurity tools,” Udeshi said.

The key innovation was developing “Interactive Agent Tools” that convert visual cybersecurity programs into text-based formats the AI can understand. Traditional cybersecurity tools like debuggers and network analyzers use graphical interfaces with clickable buttons, visual displays, and interactive elements humans can see and manipulate.

“Large language models process text only, but these interactive tools with graphical user interfaces work differently, so we had to restructure those interfaces to work with LLMs,” Udeshi said.

The team built their own dataset by collecting and structuring Capture The Flag (CTF) challenges specifically for large language models. These gamified cybersecurity competitions simulate real-world vulnerabilities and have traditionally trained human cybersecurity professionals.

“CTFs are like a gamified version of cybersecurity used in academic competitions. They’re not true cybersecurity problems that you would face in the real world, but they are very good simulations,” Udeshi said.

“We built our own CTF benchmark dataset and created a specialized data loading system to feed these challenges into the model,” said paper co-author Minghao Shao, a NYU Tandon Ph.D. student and Global Ph.D. Fellow at NYU Abu Dhabi.

The framework includes specialized prompts that provide the model with instructions tailored to cybersecurity scenarios.

Performance Against Benchmarks

EnIGMA demonstrated superior performance across multiple benchmarks. The system ended up tested on 390 CTF challenges across four different benchmarks, achieving state-of-the-art results and solving more than three times as many challenges as previous AI agents.

During the research conducted 12 months ago, “Claude 3.5 Sonnet from Anthropic was the best model, and GPT-4o was second at that time,” Udeshi said.

The research also identified a previously unknown phenomenon called “soliloquizing,” where the AI model generates hallucinated observations without actually interacting with the environment, a discovery that could have important consequences for AI safety and reliability.

Beyond this technical finding, the potential applications extend outside of academic competitions.

“If you think of an autonomous LLM agent that can solve these CTFs, that agent has substantial cybersecurity skills that you can use for other cybersecurity tasks as well,” Udeshi said.

The agent could potentially end up applied to real-world vulnerability assessment, with the ability to “try hundreds of different approaches” autonomously.

The researchers acknowledge the dual-use nature of their technology. While EnIGMA could help security professionals identify and patch vulnerabilities more efficiently, it could also potentially be misused for malicious purposes. To that end, the team notified representatives from major AI companies including Meta, Anthropic, and OpenAI about their results.

Pin It on Pinterest

Share This