The rapid adoption of artificial intelligence agents and applications is creating significant security challenges for enterprises. Niv Braun, co-founder and CEO of Noma Security, highlighted at RSAC Conference 2026 that AI systems are inherently non-deterministic, meaning they do not operate with the predictability of traditional software. This unpredictability, combined with a potentially enormous blast radius and intense pressure to deploy quickly, leaves security teams struggling to keep pace.
Core Principles for AI Security
According to Braun, effective AI security must be built on two foundational principles. First, a holistic and flexible framework is needed to absorb fast evolving technologies like the Model Context Protocol (MCP). Second, deep contextualization is critical. This involves connecting posture management, access controls, and runtime monitoring into a single, unified signal. Braun stated that without visibility into runtime behavior, it is impossible to provide accurate recommendations for configuration and access control.
Impact and Strategic Approach
Braun argued that point products are insufficient for AI security. A unified platform is necessary to understand which agent actions are legitimate versus those that represent a genuine risk. He also emphasized the importance of early partnerships between AI providers and security vendors to enable secure-by-design capabilities. Noma Security, founded by a former member of Israel’s Unit 8200, has grown rapidly since late 2024 by focusing on this integrated approach to AI and agent security.
Source: Healthcareinfosecurity
