With AI agents set to become more ubiquitous in the coming years, it might be necessary to draw distinctions between agents that are for work versus those that are for play.
Microsoft CEO Satya Nadella has envisioned a future in which there are different AI agents for the same person for their work and personal life, and with differing data access for both. “ If agents are gonna have access to data, it needs to be subject to the same data protection and data rights,” he said in an interview. “You want to manage the agent environment like an endpoint to make sure that there is no credential theft. So all of the things we’ve done for identity management and IT infrastructure security is going be done for agents and their IT infrastructure,” he added.
“I also suspect that a lot of people are going to build their own personal agents for their personal life, and maybe do you see a future in which they are also bringing those personal agents into work?” the interviewer asked him.
“That’s a great question. Yeah, the system where you bring your own personal agents has to be done in such a way that these two worlds don’t have the data leakage. Let’s even take the simple thing like my email and my corporate email. Today, they are two segregated things. There are two identities, and we know how to separate state out for both privacy reasons and for also intellectual property reasons. Both of those are helpful,” he added.
Nadella seemed to be hinting at a future where AI agents will govern different parts of our lives. The AI agent for work, for instance, could go through work emails, respond to coworkers, and even submit code. A personal AI agent could help with grocery shopping and making restaurant reservations. Even though these two agents belong to the same person, they should not have access to the same data — the personal AI agent wouldn’t need to have access to work data, and the work AI agent wouldn’t necessarily need to know where you’re eating next week. As such, it would be important to create separate data policies for these two agents, and give them access only to those parts of the data repositories that they need. This is just one of the issues that’ll come up with widespread AI agent use, but it appears that tech companies are already thinking about these use-cases as they prepare for a more AI agent-centric world.