Microsoft CEO Satya Nadella says AI agents should be treated like employees, here is why

Published 7 June 2026 · tech

As AI agents become more powerful and are trusted with increasingly important tasks, Microsoft CEO Satya Nadella believes companies need to start treating them more

As AI agents become more powerful and are trusted with increasingly important tasks, Microsoft CEO Satya Nadella believes companies need to start treating them more like employees. Speaking on the "Possible Podcast," Nadella said AI agents should be given identities, permissions, and clear rules about what they can and cannot access. Companies should also be able to audit their actions, much like they monitor the work of human employees. Read Full Story "You need to give them identities, you need to give them sandboxes, then you need to set policies to govern them," Nadella said. His comments come at a time when AI agents are rapidly moving beyond simple chatbots and becoming active participants in the workplace. AI agents are changing how people use the internet For years, AI tools were largely focused on helping people write emails, summarise documents, or improve productivity.

But the next wave of AI is centred on agents. These AI agents are autonomous, AI-powered assistants that can perform tasks on a user's behalf. An AI agent could search for the cheapest flight, book tickets, fill out forms, make reservations, and even add details to a calendar without requiring constant human input. As these agents become more capable, they are also changing how people access information and services online. Companies that were once highly protective of how users interacted with their platforms are increasingly opening their systems to AI agents. Recently, Morgan Stanley announced that it would allow AI agents to access data and insights from its platform, highlighting how businesses are adapting to this new reality. The growing risks of autonomous AI But giving AI agents more responsibility also comes with risks. There have already been reports of AI agents deleting databases, misinterpreting instructions, or carrying out actions that caused significant problems for organisations.

A single misunderstanding of a user's request, or an agent ignoring part of an instruction while completing a task, could potentially have serious consequences. That is one reason why Nadella believes governance will become critical as companies deploy more AI agents. According to the Microsoft CEO, organisations need systems that clearly define what an AI agent can access, what actions it can take, and how its behaviour can be monitored. Microsoft is already managing hundreds of AI agents Nadella said Microsoft is already dealing with these challenges internally. The CEO revealed that he often runs around 100 AI coding agents at the same time. Managing them individually through chat interfaces can be difficult. "The cognitive load on me managing this is so high," he said. To address the issue, Microsoft has developed tools that help manage AI agents at scale. Nadella pointed to Agent 365, a suite that includes Entra, Microsoft's digital identity and network access platform, as well as Purview, which helps label and track data created by AI agents.