Skip to content

Guide · AI tools

How to check whether an AI browser extension is safe

Before you install any AI or privacy browser extension, check four things: its permissions, where the checking happens, what it stores and who's behind it.

By Paulo De Sousa, Co-founder of SimpleKPI6 min read

A magnifying glass over an AI browser extension's details, with permissions, publisher and privacy policy ticked and access limited to this site only

Someone on your team installs a free "AI helper" extension on a Tuesday. It can read every page they open.

The short answer: Check four things before you leave it there: what it can read, where it checks, what it stores, and who's behind it.

If staff type a client's name or contact details into an AI tool, that's personal data. An extension that can read the page can see it too, so treat the extension as a supplier you can answer questions about. The ICO explains what counts as personal data, and what to ask of anyone who handles it for you. Articles 28 and 32 of the UK GDPR are the formal wording on processors and security.

Here's how to check, and how Saiela measures up. It's general information, not legal advice.

Four checks before you trust an AI browser extension

  1. What can it read? Only the sites its job needs.
  2. Where does it check? On your computer, not on its own servers.
  3. What does it store? A record of what happened, not what you typed.
  4. Who's behind it? A real company you can look up.

Saiela saiela.co.uk

1. What can it read? (permissions)

When you add an extension, Chrome and Edge show the permissions it asks for. Read them.

  • Access on all sites is the broadest. In these browsers it can appear as a warning that the extension can read and change all your data on all websites. It needs a very good reason.
  • Access to specific sites only is narrower. An extension that only needs to work on AI tools should only ask for those.
  • Access to browsing history, downloads or the clipboard should match what it does.

You can also change site access later, in the extension's details, to the current site or to specific sites only.

Ask the vendor why it needs each permission. A trustworthy one will say, in plain words.

2. Where does the checking happen?

This is the question most people miss.

Some AI privacy tools send what you type to their own servers to check it, and only then pass it on. A third company now sees your client data, which is the very thing you were trying to avoid.

The safer design checks everything on your own computer, so the text never leaves it except to the AI tool you chose.

Checks on a server

Your computer

What you type

The privacy tool's servers A third company sees your text

The AI tool

Checks on your computer

Your computer

What you type

Checked on your computer No third company in the middle

The AI tool

Two designs for an AI privacy tool. On the left, its servers see what you type before the AI tool does. On the right, the checking happens on your computer, so the text only goes to the AI tool you chose.

A privacy tool that uploads what you type is just another place the data went.

3. What does it store, and where?

Find out:

  • what's kept: the text you typed, or just a record of what happened
  • how long it's kept
  • where it's stored: in the UK, the EU or somewhere else
  • whether you can delete it

4. Who's behind it?

Check there's a real company you can look up:

  • a registered company name and number (in the UK, search Companies House)
  • a registered address
  • a privacy notice that names them
  • a way to contact a person

Also check the store listing

  • Reviews: read the recent ones, not just the star rating.
  • Update history: an extension that's never updated may not keep up with AI tools or security fixes.
  • Who publishes it: the publisher's name should match the company.

Microsoft's security researchers have found extensions that look like AI helpers and collect what people type into AI tools. That's another reason to check who's behind one, and to watch what it sends.

Don't take our word for it: watch what an extension sends

You can see the requests a page makes, and where they go, in the browser's Network tab. These browsers both include this in their developer tools.

  1. Open the AI tool in your browser.
  2. Open DevTools (Ctrl+Shift+J on Windows, or right-click the page and choose Inspect), then select the Network tab. It only records requests after you open it.
  3. Type a made-up test sentence, with fake names and numbers you can search for. Don't use real client details.
  4. Look at the new requests. Right-click the table header and turn on the Domain column if it isn't showing, so you can see where each one went.
  5. Use search to look for a word from your test sentence. Click a request to inspect its headers and content.

If your test sentence appears in a request to a domain that isn't the AI tool, the text is leaving the computer. It's a practical check, not a full security audit.

The same checklist, applied to Saiela

Ask all of this of us too.

Check Saiela
What can it read? What you type or paste into the AI tools it protects, before it's sent. Our extension privacy notice explains what it reads and why.
Where does checking happen? On your own computer. What you type is never sent to Saiela.
What does it store? A log of what happened (what kind of data was found, which tool, what we did), never what anyone wrote. Records are kept in a UK data region.
Who's behind it? SimpleKPI Ltd, company 06191007, 66 Paul Street, London. Building software since 2007.

There's more detail on our security page.

What to do this week

  1. List the browser extensions your team has installed. Our guide to shadow AI covers how to find tools people have added without asking.
  2. Run the four checks on any that can read what people type, and use the Network tab if you want to see the traffic yourself.
  3. Remove anything you can't get good answers for.
  4. Add the extensions you keep to your list of suppliers. You can use our free AI tool and risk register.
  5. If you manage the team's browsers, these browsers let you allow only the extensions you've checked, and block the rest.

Reviewed on 12 November 2026.

Sources

General information, not legal advice.

Questions

The short answers.

Which extension permissions are a warning sign?

Be careful with an extension that can read and change data on every website when its job only needs a few. Ask why it needs each permission. A good vendor explains them in plain words.

Are privacy extensions safe?

Some are, some aren't. An extension that sends what you type to its own servers to check it is collecting the very data it claims to protect. Look for one that does the checking on your own computer.

How can I check Saiela?

Use the same four checks: what it can read, where it checks, what it stores and who's behind it. Our extension privacy page lists what it reads and why. You can also watch what it sends in the browser's Network tab, as described in this post.

General information, not legal advice.

A laptop showing an AI chat where a file called All_Staff_Salaries.xlsx has just been sent, and the assistant replies "Well, this is awkward."

GuideData protection

Client data went into ChatGPT? What to do now

Someone put client data into ChatGPT? Don't panic: find out what was shared and on which account, then decide if the ICO needs telling. Step by step.

Stuart Kinsey3 min read

All posts

Protection that never sees your client data.

Saiela checks prompts on your computer, so we never see what your team types. There's a free plan and a 7-day Team trial.