Token Security has introduced a set of tools intended to help organisations manage the growing volume of AI agents operating across their systems. The move aims to extend the company’s existing identity governance framework to cover autonomous models, custom GPTs and services using the Model Context Protocol.
The update comes in the context of a recent increase in internal agent deployment. Moderna, for example, moved from several hundred to several thousand internally built agents within a year. While these tools support automation and efficiency, their proliferation has created large pools of unmanaged identities with access to sensitive systems, generating compliance gaps and opportunities for misuse.
Expanding governance to machine-driven identities
The company’s platform now offers automated discovery of internally developed, personal and third-party agents, mapping their activity and assigning oversight to human owners. It can also withdraw access from dormant agents to reduce the risk of long-term, unmonitored system entry points.
HiBob, which deploys AI tools within its HR technology environment, has been testing the new capabilities. According to representatives from the company, the technology provided oversight that had previously been unavailable, helping identify custom agents created by staff and ensuring that those systems operated within expected security boundaries. They said the added visibility allows them to expand AI use with fewer concerns about unmonitored access.
Token Security officials noted that conventional identity and access management systems were designed for human users and are not suited to the speed and variety of autonomous agents. Representatives said their platform helps security and IT teams apply lifecycle controls to this new category of identity, from creation to deprovisioning.
The release includes continuous discovery functions, ownership assignment, permission reduction aligned to each agent’s purpose, and audit logging across multi-agent environments. It integrates with major AI ecosystems, including platforms from OpenAI, Anthropic, AWS, Microsoft and others, allowing organisations to enforce consistent policies across varied tooling. The new features are available within the existing Token Security Platform and will be demonstrated at an upcoming industry event focused on identity and access management.
