As artificial intelligence continues its rapid integration into operational technology (OT), industrial control systems (ICS), and IoT environments, a new layer of hidden risk is emerging. Increasingly, AI models are being compiled directly into binaries and deployed on embedded devices, quietly powering automation, safety, and decision-making functions while remaining opaque to traditional security tools.
At S4x26, ObjectSecurity will present a technical session titled “Recognizing & Analyzing Compiled AI in OT Binaries & Firmware.” The session will offer practical insights into uncovering and assessing these embedded models.
The presentation will explore how compiled AI components can introduce novel attack surfaces, including risks of tampering, unauthorized reuse, and intellectual property theft. ObjectSecurity will demonstrate techniques for identifying AI artifacts within binaries and firmware images, along with practical approaches to analyze them effectively and assess their security posture.
These methods will be mapped to governance and compliance frameworks such as the NIST AI Risk Management Framework, showing how organizations can systematically evaluate and mitigate AI-related risks within OT systems. As AI becomes a core component of industrial software and embedded systems, visibility and accountability are more critical than ever. ObjectSecurity’s work in this area helps bridge the gap between AI assurance and OT security, providing a foundation for protecting the next generation of intelligent infrastructure.
