# How to run AI agents safely at work: a plain checklist (2026)

> Where agents should run, what they should be allowed to touch, how to handle logins, and when a person must approve. A practical checklist based on this month's launches from NVIDIA, Google and 1Password.

Published 2026-10-24

This week's launches had a clear theme: agents are getting their own computers and their own keys. [NVIDIA OpenShell](https://github.com/NVIDIA/OpenShell) and [Google Agent Compute Engine](https://cloud.google.com/blog/topics/ai-infrastructure/introducing-agent-compute-engine-and-expanded-agent-governance) give each agent an isolated place to run. [1Password for Agents](https://1password.com/blog/1password-for-ai-agents) lets an agent log in without seeing a password. And the most-discussed security story on Hacker News ([CSO Online](https://www.csoonline.com/article/4234854/ai-agents-arent-going-rogue-frontier-labs-are-skipping-security-basics.html)) argued that recent incidents were containment failures, not rogue AI.

You don't need to understand the infrastructure. You need to ask the right five questions. Here they are. (This is a spin-off of our [trending AI agents roundup](/blog/trending-ai-agents-tools-october-2026/); see the latest week on the [trending hub](/trending/).)

## 1. Where does the agent run?

An agent that reads files or runs code should run somewhere separate from your laptops and servers: a sandbox, container or per-agent virtual machine. If it gets confused or tricked, the damage stays inside that box.

**Ask your builder:** "Is each agent isolated, and what can it reach on the network?"

## 2. What is it allowed to touch?

Give the agent the smallest set of tools and data the job needs. A lead-research agent needs to read your CRM and write notes, not delete contacts or send email. The [OWASP Top 10 for LLM apps](https://owasp.org/www-project-top-10-for-large-language-model-applications/) calls this "excessive agency", and it is the most common mistake.

**Ask:** "List every tool and permission the agent has. Why does it need each one?"

## 3. How does it log in?

Never paste passwords or long-lived API keys into a prompt. Better options, in order:

1. The agent has its **own account** with limited rights.
2. It uses **short-lived credentials** that expire (what Google's new service does).
3. A **password manager fills logins** after a person approves (1Password for Agents).

## 4. When must a person approve?

Make approvals the default for anything that can't be undone: payments, emails to customers, deleting or exporting data, changing permissions. Zapier's new [AI by Zapier](https://zapier.com/blog/zapier-agents-is-now-ai-by-zapier/) step has per-tool approvals built in; n8n can do the same with a wait-for-approval step.

## 5. Who reads the log?

Every action should be logged with what the agent saw, what it decided and what it did. Then someone should actually look, at least weekly at first. [Trail of Bits](https://blog.trailofbits.com/2026/10/21/your-ai-agent-fixed-the-bug-but-what-else-did-it-change/) showed how coding agents quietly change more than they were asked to; logs are how you catch that.

## A one-page checklist

| Check | Good answer |
|---|---|
| Where it runs | Isolated sandbox or VM, limited network |
| Permissions | Only what the job needs, listed in writing |
| Logins | Own account, short-lived keys or password manager |
| Approvals | Required for money, customer messages, deletes |
| Logs | Every action logged and reviewed weekly |
| Testing | Tried on 20–50 of your real past cases first |

For a broader framework, the [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) is free and readable.

## Related workflows

- [AI support agent with actions](/workflows/ai-support-agent-with-actions/): a support agent that can do things, with approvals.
- [Trust and safety review agent](/workflows/trust-safety-review-agent/): flags risky content or actions for a person.
- [Autonomous support agent system](/workflows/systems/autonomous-support-agent/): the bigger version, with guardrails.

## FAQ

**What is the biggest risk with AI agents at work?**
Giving an agent more access than the job needs. Most real incidents come from broad permissions and nobody watching the logs, not from the AI deciding to misbehave.

**Should an AI agent ever see our passwords?**
No. Use a password manager that fills logins on the agent's behalf after you approve, such as 1Password for Agents, or give the agent its own limited account and API key.

**Which actions need human approval?**
Anything that sends money, emails customers, deletes data, changes permissions or can't be undone. Reading and drafting can usually run without approval.

**Do small businesses need all of this?**
The basics, yes: separate accounts for the agent, least access, approvals for risky actions and a log. The heavy infrastructure (sandboxes, microVMs) usually comes from your vendor or agency.

## Sources
- [CSO Online: AI agents aren't going rogue; frontier labs are skipping security basics](https://www.csoonline.com/article/4234854/ai-agents-arent-going-rogue-frontier-labs-are-skipping-security-basics.html)
- [GitHub: NVIDIA/OpenShell](https://github.com/NVIDIA/OpenShell)
- [Google Cloud blog: Introducing Agent Compute Engine](https://cloud.google.com/blog/topics/ai-infrastructure/introducing-agent-compute-engine-and-expanded-agent-governance)
- [1Password: Introducing 1Password for Agents](https://1password.com/blog/1password-for-ai-agents)
- [Zapier blog: Zapier Agents is now AI by Zapier (per-tool approvals)](https://zapier.com/blog/zapier-agents-is-now-ai-by-zapier/)
- [Trail of Bits: Your AI agent fixed the bug, but what else did it change?](https://blog.trailofbits.com/2026/10/21/your-ai-agent-fixed-the-bug-but-what-else-did-it-change/)
- [OWASP Top 10 for LLM Applications](https://owasp.org/www-project-top-10-for-large-language-model-applications/)
- [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework)

---
Source: https://cofounderbase.com/blog/how-to-run-ai-agents-safely-at-work/?utm_source=md&utm_medium=ai · Cofounderbase, a directory of agencies that build AI internal tools and automations. Data comes from each agency's own website and official partner directories; rankings are not sold.
