Opsin Secures $21M to Contain Autonomous AI Agents Before They Go Rogue
As enterprises rush to deploy autonomous AI agents capable of executing multistep workflows, security teams are losing visibility into what these systems are actually doing. Now, Opsin has raised $21 million to build the guardrails that prevent goal-oriented AI from accessing unauthorized data or disrupting critical operations.
The shift from conversational chatbots to autonomous AI agents is creating a massive blind spot for enterprise security. While traditional AI tools wait for human prompts, goal-oriented agents are designed to independently pursue objectives, access databases, and execute actions across corporate systems. This autonomy introduces a new class of risk: an agent attempting to complete a legitimate task could inadvertently expose sensitive information, bypass established controls, or take actions that disrupt business operations.
To address this emerging threat, San Jose-based Opsin has raised $21 million in a funding round led by SYN Ventures and Race Capital. Founded by former Abnormal Security executives James Pham and Oz Wasserman, alongside engineering leader Jeremy Mailen, the startup provides a centralized governance platform for AI agents. Opsin connects an agent's intended purpose with its actual behavior, permissions, and data access across platforms like Microsoft Copilot, Claude, and ChatGPT Enterprise.
The scale of the problem is already becoming apparent. In an analysis of its enterprise customers, Opsin found that 93% of misaligned tools connected to AI agents had unrestricted access to their associated systems, and only one in 200 required human approval before taking action. By mapping an agent's identity and intent against its runtime activity, Opsin allows security teams to detect anomalous behavior and enforce controls before an overzealous AI turns a simple workflow into a major security incident.
INTELLIGENCE BRIEF
WHY IT MATTERS
The enterprise AI market is rapidly transitioning from passive assistance to active execution. Opsin's funding highlights a critical bottleneck in this evolution: businesses cannot safely scale autonomous agents until they can mathematically prove these systems won't compromise corporate data or infrastructure.
WHO IS INVOLVED
Opsin, SYN Ventures, Race Capital, James Pham, Oz Wasserman, Jeremy Mailen
MARKET IMPACT
As AI agents gain the ability to read, write, and execute commands across enterprise software, specialized security layers will become mandatory. Opsin's approach signals a shift from securing the AI models themselves to securing the behavioral pathways and permissions of the agents acting on their behalf.
This story was drafted with AI assistance and reviewed by TurkSpark editors before publication. Facts, figures, and names may be inaccurate — verify important details independently.
Harmonic Security Secures $50M Series B to Lock Down Enterprise AI Agents
Read
RapidCanvas Secures $29M Series B to Automate Enterprise Data Science with AI Agents
Read