Onyx Security: the execution layer that enterprise AI left ungoverned

88% of organisations reported confirmed or suspected AI agent security incidents in the past year, according to the Gravitee State of AI Agent Security 2026 Report, based on a survey of 750 CTOs and technology leaders - yet only 14.4% deploy agents with full security or IT approval. The tooling enterprises rely on was designed for deterministic software operated by humans. It was not built to govern systems that act autonomously, access sensitive data, and execute decisions across platforms at machine speed.

Onyx Security, headquartered in Tel Aviv and New York, was founded by Maxim Bar Kogan (CEO), a cybersecurity leader and former member of Unit 8200 of the IDF, and Gil Elbaz (Chief AI Officer), a veteran AI researcher who previously reported to the CTO of Nvidia and served in one of the IDF's AI research units. The company emerged from stealth in March 2026 with a platform already deployed by multiple Fortune 500 companies, with a team of over 70 across Israel, the United States, and Canada.

At the core of the product is the Onyx Guardian Agent, a supervisory AI layer that continuously discovers every AI asset across the enterprise: agents, models, AI-powered applications, and MCP-connected tool ecosystems. Operating at runtime, it intercepts agent actions before they reach downstream systems, blocking unsafe behaviour, requiring human approval, or narrowing an agent's scope. The platform has secured more than 137,000 agents across enterprise deployments, analysing more than 10 million sessions for threats in real time. Operators access unified observability across SaaS, cloud, endpoint, and code environments, with governance policies expressible in natural language and enforced at the execution layer.

In March 2026, Onyx Security raised $40 million from Conviction Partners and Cyberstarts.

Sources: Onyx Security | Gravitee State of AI Agent Security 2026 | Jerusalem Post

Founders: Maxim Bar Kogan, Gil Elbaz