Okta CEO Todd McKinnon called for companies deploying artificial intelligence agents to build kill switches into their systems to instantly revoke access if the technology goes rogue. The cybersecurity company released a framework on March 15, 2026, detailing security measures for businesses using AI agents in their operations.

McKinnon described AI agents as a new class of digital workers that can access systems, move data, and take actions across a company's software stack. He emphasized that companies need a system to track AI agents, define their roles, permissions, and connectivity, with the ability to revoke an agent's access if it goes rogue. "It's almost like you would take a machine off the network," McKinnon said.

Okta published a press release titled "The blueprint for the secure agentic enterprise" that recommended AI-using firms be able to revoke access instantly across every system to contain risk. The framework calls for real-time enforcement of data-sharing permissions, human approval for risky actions by AI agents, and detailed audit logs that track every agent's decision and access attempt.

Businesses are increasingly experimenting with AI agents to automate workflows, build software, and handle some physical tasks. "But for agents to really do their job, they need access to sensitive systems and data, thereby creating a new attack vector," said Harish Pari, senior vice president of AI security at Okta. Okta is positioning itself as the security layer for AI agents as companies adopt the technology.

In early 2024, California State Senator Scott Wiener proposed an AI regulation bill requiring firms to build a failsafe into their AI systems. The bill received support from Elon Musk but was vetoed by Governor Gavin Newsom later in 2024.

Every day, millions of workers use Okta to gain access to applications and other services. McKinnon leads the company as its CEO.