Vero Security raises $11M seed to turn AI security policy into runtime controls that follow vendor changes
Synopsis
Vero Security announced an $11 million seed round led by TLV Partners, with GTM Capital, Maple Capital and angel investors participating. It targets the AI security work that enterprises often assign to someone who already holds another role, turning security-team policy into enforced controls, adjusting them as AI vendors change their tools, and bringing back only decisions that need human judgment. The company says multiple Fortune 500 companies already use it, and it launched AISecurityEngineering.org, an open community for practitioners in the field.
Interpretation
The people responsible for AI security at enterprises usually hold another job, and what they must protect keeps moving: employees adopt new AI tools and agents, connect them to enterprise systems and grant permissions to reach sensitive data and take actions, while AI vendors keep adding capabilities, models, plugins and integrations that can change the security meaning of existing configurations. Research from IANS and Artico Search found that 65% of companies with at least $10 billion in revenue had assigned someone to lead AI security, but only 18% had made it a dedicated role; the rest added it to an existing leader's job. Vero says a new Fortune 1000 customer started with one policy, allowing only developers to connect AI agents to corporate code repositories or let them execute code, and Vero turned that single decision into more than 15 security controls across coding agents, MCP servers, skills, identities and runtime actions within minutes, enforcing them organization-wide.
The platform centers on an AI Runtime Governor that puts security decisions into effect while AI operates, rather than waiting for manual security review. Runtime controls inspect supported prompts, tool calls and agent actions and enforce approved policies at the moment an action is attempted, so blocking happens during execution instead of after a review queue. By the company's description, it can stop a coding agent from reading secrets it has no reason to touch, block agent actions that violate security policies, and keep malicious skills or prompt injection from triggering unauthorized behavior; it works through the endpoint, identity, AI and security tools a company already runs, so employees install nothing new.
Perspective
The report covers Vero Security's funding announcement, its description of platform functions and the customer deployment the company recounts, for researchers and engineers following enterprise AI security operations.
The company does not disclose customer counts, deployment scale or measured interception results; runtime controls cover only supported prompts, tool calls and agent actions, and how other cases are handled is not stated.
