Onyx Security sells what it calls a secure AI control plane for enterprise AI agents. The platform discovers sanctioned and shadow AI across SaaS, cloud, endpoints and code, and gives visibility into prompts, responses and agent actions. It enforces security and governance policies, which teams can define in natural language, and a supervisory Guardian Agent monitors the platform and can block an action, require human approval or redirect the agent. Listed integrations include Anthropic, OpenAI, Amazon Bedrock Agents, Microsoft Copilot Studio and Google. It is delivered as a vendor-operated platform.
Onyx Security came out of stealth in March 2026 and lists offices in Tel Aviv, Israel, and New York. Israel is outside the EU and EEA. The company operates a Trust Center that lists compliance certifications and subprocessors on request. The sources consulted do not state where the platform is hosted and describe no self-hosted option or open-source edition, so European organisations should ask where their data is processed.