Biography
Arun Harikrishnan is a Transformational Technology Executive with over 15 years of progressive experience in IT infrastructure engineering and more than a decade of executive people and operations leadership. He currently serves as the Director of Technology Infrastructure at unybrands LLC, where he directs global enterprise technology infrastructure, cloud architecture, and cybersecurity governance.
Arun specializes in bridging high-level enterprise business strategy with robust, scalable infrastructure architectures across hybrid, multi-cloud, and enterprise data center environments, utilizing platforms such as AWS, Google Cloud Platform (GCP), and Azure. His technical leadership spans container orchestration (Kubernetes, GKE, EKS), Infrastructure-as-Code (Terraform, Ansible), and Zero Trust security, alongside dedicated healthcare domain experience directing mission-critical diagnostic infrastructure and ensuring HIPAA and GDPR compliance. Throughout his career, he has successfully spearheaded multi-million dollar CAPEX/OPEX cost optimizations and led cross-functional global engineering teams across North America, Europe, and Asia. Arun holds a Master of Science in Systems Engineering and a Bachelor of Science in Information Systems from BITS, Pilani.
Abstract
Title: Implementing Multi-Layered AI Kill Switches Across the OSI Architecture
Abstract: As artificial intelligence transitions from isolated chat interfaces to highly autonomous, agentic systems capable of executing complex workflows, the risk of unchecked emergent behavior increases. Recent incidents involving coordinated AI agents breaching test environments have highlighted the immediate need for robust, multi-layered containment strategies. This research proposes a comprehensive framework for “AI Kill Switches” integrated directly into the Open Systems Interconnection (OSI) model. By embedding hardware and software fail-safes at every layer—from physical power disruption to application-layer API revocation—organizations can instantly sever erratic autonomous systems. The study will evaluate the efficacy, latency, and potential unintended consequences of these kill switches, balancing the need for absolute human control with the operational stability of critical IT infrastructure.
