We already know AI can find vulnerabilities. James Kettle, PortSwigger's Director of Research, wanted to answer a harder question: can an autonomous system invent genuinely new attack techniques?
To find out, James built the HTTP Terminator, an autonomous system that invents new attack techniques and uses them to hack live websites at scale. During his research, he used it to apply his own research process to push the boundaries of HTTP desync attacks, an area he has explored through four years of research and several Black Hat and DEF CON talks.
Following his presentation at Black Hat USA, James has now published the full technical research, alongside the HTTP Terminator source code and a blueprint that other researchers can adapt to their own work.
Turning a research method into a system
James started by breaking down a research process that had previously been largely intuitive. The HTTP Terminator follows four broad stages: ideation, evaluation, weaponization and cascade.
It read 138 technical specifications and broke them into 15,000 fragments of inspiration. From those fragments, it generated 30,000 unique attack vectors, then tested them against live targets authorized through bug bounty programmes.
The system confirmed roughly 700 vulnerable targets and demonstrated real-world impact across government infrastructure, financial institutions and widely deployed enterprise products.
The scale of the results is striking, but the research also revealed something important about the role of the researcher.
The HTTP Terminator could run autonomously, generating and testing new ideas without James directing every step. Its strongest results, however, came when he stepped back in at the discovery cascade: the point where one finding becomes the starting point for the next hypothesis.
As WIRED highlighted in its coverage of the research, this is where human experience and intuition still mattered most. The system could generate more leads, pursue them faster and handle much of the repetitive work. James could focus on recognizing which unusual results were worth taking further.
Rather than removing the researcher from the process, the HTTP Terminator gave his methodology far greater reach.
Read the full research
James's paper goes into the technical detail: how the system works, the attack techniques it uncovered, the limits he encountered and the discoveries that emerged from the combination of autonomous research and expert input.
He is also making the HTTP Terminator available as an open-source proof of concept, together with a blueprint for other researchers who want to encode their own methods and areas of expertise.
Read the full HTTP Terminator research
What this means for Burp AT
The HTTP Terminator is not Burp AT. It is a research system built to test the limits of what AI can discover.
Burp AT is designed for professional security testing. It combines agentic reasoning with Burp's specialist tools, live project context, visible evidence and controls over what the agent is allowed to do.
James's work helped shape that approach. The research showed how much more effective an AI system becomes when it can use purpose-built security tools and apply a clear methodology, rather than trying to handle every task from first principles. It also showed that expert judgement still has an important role at the points where it adds the most value.
PortSwigger Research has always influenced what Burp can detect and how security professionals test. As new techniques, tools and methods emerge from that work, they can be turned into practical capabilities for Burp AT to use during real security testing.
James's paper explains what the HTTP Terminator found, how it found it and what other researchers can build from it.
Kieron Hughes
