AI security for entrepreneurs

Use AI in your business without exposing it.

AI saves you hours because it can get into your files, your email and your accounts. That same access is what can take information out of your business without anyone noticing. These are the four things worth checking today.

Write to us See the 4 things to check

Why this matters now

Until recently, security was a problem for big companies with IT departments. Today anyone with a phone uses AI to write, summarise and automate. That's good. The problem is that nobody explained what happens to the information you give it, or what you're allowing when you click "connect".

How AI can take information out of your business without you noticing

You asked for help with one file. Now it can see all of them.

The risk: To summarise a contract, you connect AI to your Google Drive or OneDrive. But the permission is almost never for that one file: it's for everything. Photos of your clients' ID documents, the payroll sheet, contracts from three years ago. And deleting the app from your phone doesn't take that access away.
What to do today: If you only need help with one file, upload that file instead of connecting your whole account. If an app asks for your entire Drive for a small task, don't connect it. And open the connected-apps section of your Google or Microsoft account to remove anything you no longer use.

An agent also obeys what it reads.

The risk: If an AI assistant reads your emails, documents or web pages for you, any text it reads can give it orders. An email with a hidden instruction, in white text on a white background, can ask it to find an invoice and send it to someone. You don't see the instruction, but the agent does. The chatbot that answers your customers works the same way: a well-written message can get it to show what it shouldn't.
What to do today: Don't give one assistant these three things at once: reading messages from strangers, access to your files, and the ability to send things on your behalf. If it needs all three, nothing goes out without your approval.

"Share" can mean "anyone with the link".

The risk: You ask the assistant to pass the report to your accountant, and it does it the easiest way: a link anyone can open. That link ends up in a forwarded email, a WhatsApp group or a search engine. In 2025, thousands of shared ChatGPT conversations showed up in Google because their owners ticked a box without understanding what it did.
What to do today: Check which files in your Drive or OneDrive are shared with an open link and change them to specific people. Do the same with the shared links of your AI chats.

What an agent does, it does in your name.

The risk: An agent connected to your email sends from your account, and an automation connected to your Drive moves files with your permissions. No notification arrives, because as far as the system knows, it was you. Often you find out when a client replies to an email you never wrote. And if someone on your team built the automation with their own account, it keeps running after that person stops working with you.
What to do today: Make a list of which assistants and automations have access to your accounts, and who created them. Check your sent folder and your Drive's recent activity every week. When someone leaves the team, switch off what they created the same day.

The free guide covers seven ideas, each with a concrete action, plus a 5-minute test.

Download the free guide (PDF)

When you need more than a checklist

If AI is already part of how your business runs — chatbots answering clients, automations connected to your email, a team using several tools — we review it with you and leave you with something concrete.

Which AI tools your business uses, and what information each one receives.
Which apps and automations have access to your email, files and accounts, and which ones shouldn't.
What your agents and automations do in your name, who created them, and what happens when that person leaves.

We don't sell courses or packaged products. We prepare each review for the business that asks for it.

Who's behind this

I'm Diego Rojas, a systems engineer with ten years in software, cloud infrastructure and security. Today I work on non-human identity and AI security: what software, automations and AI agents are allowed to see and do. I co-founded Stratega Studio with Jimena Alzamora, and on Instagram I explain AI security without technical jargon.

Follow on Instagram Connect on LinkedIn

Using AI in your business and not sure it's safe?

Tell us what tools you use and what you use them for. One of us will reply personally — and if a checklist is all you need, we'll tell you.

Write to us

Or send a direct message on Instagram.