Who is your AI agent when it logs in?

An AI assistant that only drafts text is one thing. An AI agent that can look up a customer, change a record, or send a message is another. The moment it can act, it needs an identity, just like an employee does.

What is changing

In late October, the US standards agency NIST published ten proposed design tenets for how AI agents should be identified, authenticated, and given permission to act. They are a proposal, not a rule. But they show where expectations are heading, and customers, auditors, and insurers tend to follow.

Why it matters for a small or mid-sized business

The easy way to set up an agent is to give it a shared admin login or someone’s password. It works on day one. It also means that if the agent is tricked, makes a mistake, or is misused, it can reach everything that login can reach, and your logs cannot tell the agent from a person.

One next step: ask four questions

  • Does each AI agent have its own login, separate from any person?
  • Does it have only the access its job needs, and nothing more?
  • Is there a log of what it did, and who reviews it?
  • Can you switch it off quickly if something goes wrong?

If you cannot answer yes to all four, fix that before the agent goes live. That is the work behind our agent identity and access review, part of AI and IT Security Consulting. If you want a second pair of eyes, get in touch.