Microsoft announces controls to limit AI agents: access only to what they need, for as long as the task lasts
Redacción IvokaOct 20, 3:00 PM 3 min
Microsoft 365 E7 and Project Perception add dynamic permissions and risk scoring for agents. If an agent looks risky, it can lose permissions or be stopped. Microsoft did not say when they will be available.
Microsoft announced new tools to control AI agents inside Microsoft 365 E7 and its Project Perception platform. The core idea: an agent starts with no access of its own and gets only what it needs for one task, for as long as that task lasts.
What is included
- Dynamic Permissions: permissions granted per task that never exceed what already exists.
- Dynamic Agent Risk: a continuous risk rating for each agent, using signals from identity, data, devices, cloud, network and the agent's own actions.
- Stepped response: depending on risk, an agent can lose the right to change files but keep read access, lose access to some data, need human approval, or be stopped.
- Guardian Agents and Playbooks in Project Perception, to coordinate protective actions.
A Microsoft example: if an expense-processing agent receives suspicious instructions and tries something unrelated to its job, the system could remove its permission to approve expenses while keeping its ability to review them.
Availability
Microsoft did not say when the new features will be available, or whether they will cost extra on top of a Microsoft 365 E7 licence.
