Silverfort brings runtime identity controls to Microsoft Copilot Studio agents
Identity security company Silverfort Inc. today launched an integration that applies its identity and access controls to artificial intelligence agents built into Microsoft Corp.’s Copilot Studio, enforcing policy at the moment an agent tries to act rather than after the fact.
The integration evaluates every access request a Copilot agent makes in real time and returns a decision before the action runs. Silverfort says this stops an agent from gaining access it should not have or escalating its own privileges before either can happen. The pitch is that enterprises can run agentic AI without handing it free rein.
A Copilot Studio agent can log in on its own, reach into enterprise data and kick off workflows across both cloud and on-premises systems. Each action traces back to a human user with a given privilege level along with multiple machine identities, creating a chain of authentications and authorizations where privilege can escalate. Microsoft reports that more than 80% of the Fortune 500 are deploying agents built with low-code and no-code tools, while 29% of employees already use unsanctioned AI agents at work.
