Breaking and Pwning LLMs: Red Teaming the Mind of Machines
by Nalinikanth Meesala & Farooq Mohammad
Nalinikanth Meesala
Cyber Security Regional Practitioner - EMEI
Nal is a Cyber Security Regional practitioner - EMEI with ~12 Years of experience. Nal's areas of expertise include AI security, Threat modeling, Threat identification, vulnerability management, assessments, building security capabilities, security automation, and DevSecOps.
Outside work, Nal is an avid painter and sportsman.
Farooq Mohammad
Cybersecurity Professional, Thoughtworks
Farooq Mohammad is a cybersecurity professional at Thoughtworks with over a decade of experience in application and mobile security. He works closely with developers, infra teams, and business stakeholders to build secure, scalable systems, with a strong focus on DevSecOps. Lately, he's been diving deep into the intersection of AI and security. Farooq loves breaking things (ethically) and helping teams shift security left—without slowing down the flow.
Abstract
As LLMs like ChatGPT, Claude, and Gemini take over chatbots, copilots, and decision engines, attackers see a new playground. This talk unpacks how prompt injection, data poisoning, jailbreaks, and covert channel exploitation are turning LLMs into a new attack surface. We will also deep dive into how to red team AI systems, safely simulate attacks, and what defenses actually work.