Aembit Adds Controls for Personal AI Agents

Aembit now lets enterprises govern personal AI agents with blended identities and short-lived access at no extra cost.

Aembit said on 2026-10-06 that it is extending access controls to personal AI agents employees bring to work, at no extra cost and available immediately to existing customers. The company names Meta Muse and OpenAI Dots and says those agents can reach email, cloud apps, code repositories, and internal tools through employee credentials held outside security's control. Its model blends a distinct agent identity with the employee, evaluates policy at runtime, and issues short-lived scoped credentials rather than standing secrets. Teams can centralize audit logs and revoke a single agent without locking the user out. The same coverage is described for Microsoft Copilot Studio, Claude, ChatGPT, Gemini, and custom LLM workflows, enforced via the Aembit MCP Identity Gateway or existing gateways. The two reports agree on the product scope, identity model, and named platforms.

• On 2026-10-06 Aembit announced immediate, no-extra-cost support so enterprises can control personal AI agents that employees bring into the workplace.
• Named personal agents include Meta Muse and OpenAI Dots; the same model also covers Microsoft Copilot Studio, Claude, ChatGPT, Gemini, and custom LLM workflows.
• Those agents can reach email, cloud apps, code repositories, and internal tools using employee credentials stored outside security's control.
• Aembit assigns each agent a distinct identity blended with the employee, evaluates policy at request time, and issues short-lived scoped credentials instead of standing secrets.
• Security teams can centralize audit logs and revoke one agent without disabling or locking out the employee's account.
• Enforcement is through the Aembit MCP Identity Gateway or existing gateways, and the capability is available immediately to existing Aembit customers.