AI Agent Security Startup AIR Raises $50 Million

New AI security startup AIR has publicly launched with a substantial $50 million in seed funding, aiming to address emerging supply chain risks associated with enterprise-grade AI agents. The company was founded by Yair Saban and Niv Hoffman, both veterans of Israel’s elite Unit 8200 intelligence corps.

The funding rounds were led by venture firms Sequoia Capital (initial $10 million) and Greenoaks Capital ($40 million), with additional investment from enterprise security founders and angel investors. This financial backing underscores growing concerns about the security implications of rapidly expanding AI agent adoption across industries.

The Growing Security Challenge

As companies deploy AI agents to automate tasks, these systems increasingly rely on third-party components like plug-ins, Model Context Protocol (MCP) servers, and digital “skills” to connect with external resources. This creates new attack vectors where malicious actors could compromise agent functionality by injecting poisoned data or code.

AIR’s platform addresses this challenge by providing visibility into active agents across networks, identifying unauthorized tools, continuously monitoring components for security vulnerabilities, and enforcing compliance policies. The company claims its system currently blocks around 27% of evaluated online agent add-ons and skills.

Early Traction in Regulated Industries

Customer adoption has been particularly strong in financial services and pharmaceuticals—heavily regulated sectors where data security is paramount. AIR currently serves over 20 corporate clients, with a significant portion representing large enterprise organizations.

CEO Yair Saban emphasized the parallels between current AI agent security challenges and past software vulnerabilities: “Just like we now require digital signatures for drivers before installation, we need similar safeguards for AI skills and plug-ins to prevent unauthorized code from accessing critical systems.”

The funding comes as financial institutions face a governance gap with autonomous systems—a challenge highlighted in reports showing traditional cybersecurity infrastructure is not designed to protect dynamic AI agents that operate across software boundaries.