The growth of agentic AI has made one point clear: we’re done with role-based access control. Static, role-based permissions simply don’t map to a world where agents — autonomous AI systems that perform tasks on behalf of people or applications — become “users” and take on tasks that don’t fit clean job descriptions. It’s a trend that’s accelerating fast, with AI agent adoption surging across enterprises, and much of it happening outside formal enterprise visibility and control mechanisms. Attackers are already experimenting with ways to compromise agents, hijack their instructions, and execute full kill chains — and fixed privileges make this infinitely easier for them.

Full commentary : The growth of agentic AI has made one point clear: we’re done with role-based access control.