What does it actually take to build an AI system capable of finding real vulnerabilities? In this episode of Builders&Breakers, STÖK and KUGG sit down with Hamid Kashfi. Security researcher at XBOW and DarkCell, formerly Trail of Bits and Immunity, to explore how hackbots are changing offensive security. We discuss why the model may represent only 20% of a capable system, while the other 80% comes from the harness around it: context, memory, tools, monitoring, traces, evaluations and human expertise. Hamid takes us inside the difference between a compelling demo and a system that can reliably understand a target, ask the right questions and produce valuable findings in the real world. In this episode: • Why the model is only one part of the capability • What an effective offensive-security harness requires • The importance of context, memory and observability • Why deep findings begin with deep understanding • How AI changes the role of the security researcher • What the next generation of hackers could look like Builders&Breakers explores high-velocity shortcuts for the people building and breaking the software of tomorrow. Guest: Hamid Kashfi https://x.com/hkashfi A big thank you goes out to this episodes sponsor and research partner: Zenity Labs https://labs.zenity.io/ Subscribe for future conversations with security researchers, builders and technical leaders working at the edge of AI and cybersecurity. buildersandbreakers.com

Local AI for Offensive Security: Models, Harnesses & Human Judgment Hamid from XBOW ended our last episode with a wish for better local models. So we invited Thomas Olofsson, aka Skjortan, to tell us what he can already do with them. Thomas brings his own models into sensitive client environments and uses them to help find, verify and reproduce vulnerabilities without sending the client’s data to a cloud model. Together with STÖK and KUGG, he gets into the hardware, the specialist agents and the skills that make smaller models useful for real penetration tests. The conversation goes beyond finding bugs. They discuss why a threat model gives agents a meaningful target, why an experienced tester still needs to guide the work, and why the next big challenge may be building models that can fix vulnerabilities as quickly as others can find them. Thomas is the founder of SEC-T. https://sec-t.org/ You can find him on LinkedIn. https://se.linkedin.com/in/tolofsson Builders & Breakers is supported by Zenity Labs. https://labs.zenity.io Check out https://buildersandbreakers.com for more high velocity shortcuts