Aembit, an identity control plane for AI agents, has launched new security features for personal AI agents operating within enterprise environments. The update is available immediately to current customers at no extra charge.
The platform addresses risks associated with agents like Meta Muse and OpenAI Dots accessing corporate resources such as email, cloud applications, and code repositories. Aembit uses a blended identity approach, combining the agent's unique identity with that of the employee.
This system allows organizations to issue short-lived credentials, record agent activity, and revoke agent access without disabling the user's account. Policies can be enforced via the Aembit MCP Identity Gateway or existing infrastructure.
Beyond Meta Muse, the platform supports security for Microsoft Copilot Studio, Claude, ChatGPT, Gemini, and custom LLM workflows. The company aims to centralize access management for AI workloads to mitigate data exposure risks.
