How big of a risk are popular Chrome extensions?

I'm about to switch to Chromium and I installed a couple of extensions. Each time I installed an extension I was notified of which data the extension has access to, e.g.:

I do understand that access to that data is necessary for the extension to work, yet I'm a bit worried that such an extension might one day decide to update and steal ("phone home") all my browsing data.

Another example of a scary message (when enabling extensions for incognito windows):

Warning: Chromium cannot prevent extensions from recording your browsing history. To disable this extension in incognito mode, unselect this option.

Is that a possible threat when using popular Chrome extensions? It's a bit scary to have to trust another party for each new function you add to the browser.

Solution 1:

You are forgetting the following:

The more popular an extension is, the smaller is the chance of nobody noticing that the add-on does something harmful.

In contrast to that, if you install some extension that nobody else has used before, you risk more than, let's say, installing AdBlock. Considering that so many people are using it, it is almost safe to say: Someone would have noticed unusual traffic.

In fact, all extensions disclose their source code, so anybody could basically go ahead and look for anything suspicious themselves.

The warnings are just there so you can't blame the browser vendors for any damage done, in case you install something that goes rogue with your data. Always read the reviews of add-ons that look suspect to you before installing them.

Also note that, for example Google can check the submissions:

While Google is not obligated to monitor the Products or their content, Google may at any time review or test your Products and their source code for compliance with this Agreement, the Google Chrome Web Store Program Policies, and any other applicable terms, obligations, laws, or regulations, and may use automated means to conduct such review

Removal of an extension can of course cause some trouble to the developer.

Solution 2:

It's a tough risk-assessment to try to do. Popularity brings two things:

• More people trying to improve it (spotting bad code)
• More people trying to hack it (and introduce bad code) to attack a bigger userbase

Let's assume for these examples we're talking about an open source project with code hosted in something like github.

If something has one developer, that's just one person checking in code. If somebody (not the developer) wants to add code to that, they either need to fool the developer into adding a malicious patch (it happens), or target that developer's authentication so they can add the code themselves (also happens). The chance of either of these happening depends on the ability of the developer and their security.

If there are 10 developers, there are 10 times as many attack vectors. But also 10 times as many people that might spot the code.

I'm sure there's a point in a project where it gains enough momentum to have people performing regular security audits on its code. But at any time before that, it's swings and roundabouts.

tl;dr There are too hard to realistically work out. There are too many human elements. If it matters, don't trust it unless you can verify the code yourself.