The $40 million funding round raised by Onyx Security marks more than the launch of another cybersecurity startup. It signals the emergence of an entirely new security category centered on controlling and monitoring autonomous AI agents operating inside enterprise systems. As organizations increasingly deploy AI copilots, automation bots, and decision-making agents across their infrastructure, traditional cybersecurity frameworks—designed for networks, endpoints, and human users—are proving insufficient for this new environment.
Onyx Security was launched with $40 million in funding led by Conviction Partners and Cyberstarts, with participation from prominent cybersecurity and AI investors. The company operates between Tel Aviv and New York and brings together a founding team with backgrounds in elite Israeli cyber intelligence units and advanced AI research environments. This combination of cybersecurity expertise and machine learning experience reflects a broader trend in the industry: the convergence of AI infrastructure and cyber defense.
The company’s platform focuses on what it describes as a secure control plane for AI agents. Instead of monitoring only traditional assets such as servers or endpoints, the system is designed to detect, observe, and govern AI agents that interact with enterprise software systems. These agents can access databases, interact with APIs, execute workflows, and make operational decisions across corporate systems. As AI automation spreads through organizations, these agents effectively become a new class of digital actors inside the enterprise.
This shift introduces an entirely new attack surface. An AI agent compromised through prompt manipulation, model exploitation, or unauthorized integration could potentially execute high-impact actions at machine speed. Unlike human users, these agents operate continuously and can interact with multiple systems simultaneously. The risk profile begins to resemble automated insiders capable of performing thousands of operations in minutes.
Onyx’s architecture addresses this problem by creating a supervisory layer that continuously analyzes how AI agents reason, what data they access, and what actions they attempt to execute. The platform can intervene in real time—blocking actions, requiring human approval, or redirecting agents when behavior deviates from defined policy. In effect, it acts as a governance framework for autonomous digital workers.
The emergence of this category reflects a deeper transformation underway in enterprise technology. Over the past two decades cybersecurity evolved through successive defensive layers: first network security, then endpoint protection, followed by identity management and cloud security posture management. AI agents represent the next logical frontier. As organizations deploy thousands of autonomous processes across cloud environments and SaaS systems, security models must shift from protecting machines to supervising machine behavior.
Investor interest in the sector suggests strong belief that AI governance will become a critical part of enterprise infrastructure. Venture capital funding in cybersecurity remains resilient even during periods when other technology sectors slow. The reason is simple: cyber threats continue to evolve faster than corporate defenses, and organizations cannot afford to underinvest in protection. Platforms that promise automation, visibility, and centralized control over complex environments attract significant investor attention.
Onyx Security also reflects the continued strength of Israel’s cybersecurity ecosystem. The country has produced a disproportionate number of global cyber companies, largely due to the pipeline of talent emerging from intelligence units such as Unit 8200 and the close interaction between venture capital and defense-driven technology development. Even during periods of geopolitical uncertainty, global investors continue to fund Israeli cyber startups at high levels, betting on their technical depth and track record of successful exits.
From an industry analyst’s perspective, the central question is whether AI agent security will become a standalone market or eventually be absorbed into larger cybersecurity platforms. History suggests that both outcomes are possible. Many cybersecurity categories begin as specialized startups before being integrated into the product suites of major vendors such as Palo Alto Networks, CrowdStrike, Microsoft, or Cisco.
However, if enterprise AI adoption continues to accelerate—as current trends suggest—the scale of the problem could justify a new layer in the cybersecurity stack. Organizations may soon operate hundreds or even thousands of AI agents performing tasks ranging from software development to financial operations and customer support. Managing the behavior of these autonomous systems may become as essential as managing human identities today.
Seen through that lens, the $40 million raised by Onyx Security represents more than startup capital. It represents early investment in the infrastructure needed to secure a future where machines increasingly act on behalf of humans. In that world, cybersecurity will no longer focus only on defending systems from attackers but also on governing the behavior of the intelligent systems organizations themselves deploy.