New Free guide every week. Subscribe to never miss one. See all free guides
AI Security

Check these three settings before you connect AI to anything.

Most founders skip straight to connecting AI to their inbox, their documents, and their business tools. The settings that protect you from exposure come first. Here are the three you need to review before you plug anything in.

Connecting AI to your business tools is one of the most powerful moves you can make as a founder. When your AI system has access to your email, your documents, and your project management platforms, it stops being a chatbot and starts functioning like an actual operator in your business.

But access cuts both ways. The same connections that let AI help you draft emails and update documents can also expose sensitive information if the underlying settings are not set up correctly. Most people skip this part entirely because the tools make connecting feel easy and instant. The setup screens are clean, the integrations are quick, and nothing raises a red flag in the moment.

Before you connect AI to anything you care about, three settings deserve your attention. They take about ten minutes to review and they address the vast majority of common risks. Think of this as the foundation layer of your AI system, the part that makes everything else safer to build on.

Before you start, here's what you'll need

The setup.

Step 01 Review your privacy settings

Head to Settings, then Privacy inside your AI tool. This is where you control what the platform does with your data, not just what the AI does in the moment.

Two things to look for and turn off immediately:

  • Location sharing. Turn this off unless you have a specific reason to need it.
  • Conversation data usage. Look for any setting that allows your chats to be used for product improvement or model training and disable it. When this is on, the content of your conversations, including business information, client details, and sensitive context, can be used to train future AI models.

Not every AI tool has these settings in the same place, and some free-tier accounts have fewer options than paid ones. If you are on a free plan and the data usage settings are locked, that is worth factoring into your decision about whether to connect sensitive business tools at all.

Step 02 Understand what permissions your connectors have

Go to Settings, then Connectors (sometimes called Integrations or Connected Apps depending on your tool). Look at every connection you have active and identify what level of access each one holds.

Most connectors fall into one of three permission levels:

  • Read-only access. The AI can view and retrieve information but cannot make any changes. Lower risk. Think of it as the AI looking through your files without touching anything.
  • Interactive access. The AI can interact with the tool in limited ways such as searching or navigating. Moderate risk depending on what interactions are included.
  • Write and delete access. The AI can create, edit, send, or remove content on your behalf. Highest risk. Requires the most careful permission settings.

For each connector, you can usually choose between Always Allow, Ask First, and Block Access. Read-only connectors on Always Allow are generally fine. Anything with write or delete capabilities deserves a closer look before you leave it on Always Allow.

Step 03 Add approval requirements for anything that can make changes

This is where most founders should be more cautious than they are by default. If a connector has the ability to create, edit, send, or delete content in tools like Gmail, Google Docs, Notion, or any project management platform, set those specific permissions to Require Approval.

Require Approval means the AI has to stop and ask you before it acts. It cannot take any of the following actions until you review and confirm:

  • Send an email
  • Edit or create a document
  • Update a database or project board
  • Delete or remove any information

You are placing a human checkpoint between the AI and your business systems.

This is not about distrust. It is about building a system where mistakes, misunderstandings, or unintended actions are caught before they happen rather than cleaned up after. As you build confidence with a specific connector and understand exactly what it does, you can revisit whether Require Approval still makes sense for that particular action.

That's it.

These three settings will not eliminate every possible risk, but they address the foundation. With privacy controls locked down, permissions mapped, and approval requirements on anything that can take action, you have a much safer base to build from before connecting AI to the tools you use every day.

From here, you can start connecting with intention, knowing exactly what access you have granted, what requires your sign-off, and what stays off-limits. That is how a real AI system gets built.

Free Download

Save this guide as a PDF.

Your future self will thank you for saving this.

Want more?

Get every new free guide in your inbox.

One new free guide. Every Tuesday. No spam, no fluff, just the systems and prompts I'm building for clients, simplified for you.

Subscribe

The FREE Her AI Systems Summit drops Summer 2026. Be the first to know when dates go live. Join the Waitlist →