This post is also available in:
עברית (Hebrew)
Israeli cloud-security startup Sweet Security has secured a $75 million Series B round, and with it introduced a broadened platform designed to surface the growing number of hidden risks emerging from enterprise AI adoption.
Organizations moving rapidly to cloud-first and AI-driven architectures face a widening gap between what their systems actually run and what their security teams can fully observe. As AI models, agents, and automated processes multiply across environments, many operate without formal oversight or documentation. This creates what many security teams now describe as “blind spots” — areas where misconfigurations, excessive permissions, or unmonitored integrations can unintentionally expose sensitive data or enable unauthorized system access.
The company’s platform aims to close that gap by providing a real-time view of what is happening across cloud workloads and AI components. Its cloud-native application protection platform monitors running code, configuration weaknesses, identity use, and API behavior in a single environment. Instead of prioritizing theoretical vulnerabilities, the system correlates live activity to highlight only issues that pose an immediate operational risk. Security teams can track how identities – human and machine – behave, whether they are being misused, and which attack paths are actually viable.
The newer AI Security Platform adds a dedicated layer for monitoring AI assets. It automatically identifies every model, agent, and AI-powered service in an organization, maps how they interact, and flags unnecessary privileges or unusual communication patterns. A key use case is detecting “Shadow AI,” where unofficial or forgotten AI tools run in production without proper review. The platform also helps teams understand how agents make decisions and what data they access, providing a clearer picture of the organization’s true exposure.
Although the company is positioned as a commercial cloud-security vendor, the relevance to defense and homeland security environments is straightforward. Military and national infrastructure operators are integrating AI into command systems, intelligence workflows, and automated decision tools. Each of these components introduces new pathways for adversaries to exploit. Continuous visibility into AI behavior and cloud runtime activity is becoming essential not only for enterprise resilience but also for operational security in defense-aligned systems.
With the new funding, the company plans to accelerate product development and broaden global deployment, responding to what it describes as rising demand for practical, real-time oversight of cloud and AI environments.
