Meet Sentinel

the artificial intelligence ad detector.

With your help, Sentinel could be the future of ad blocking.

Sentinel uses machine learning to detect Facebook ads visually. The more Facebook screenshots you submit, the faster Sentinel will learn.

Team up with Sentinel for the future of ad blocking!


Here's how you can help

Step 1

Take a screenshot of your Facebook feed that contains an ad. (If you use Adblock Plus, you’ll need to disable it temporarily then reload the page to get ads)

Step 2

Send the screenshot to Sentinel with the messenger bot

Step 3

Spread the word - the more ads Sentinel eats, the faster it learns



Using a convolutional neural network we aim to visually detect ads. Read the whitepaper and check the GitHub repository for more information.

Frequently Asked Questions

Sentinel is learning how to detect ads algorithmically. Pure A.I. ad blocking.
Nothing, but it could be more efficient. Currently, ads are blocked by using huge filter lists maintained by third parties -- those lists provide instructions to your ad blocker every day to block ads contained on the lists. Also, websites like Facebook use circumvention technology to disrespect ad-blocking users. Technology helps us fight back. Sentinel uses machine learning to detect Facebook ads visually. leaving ads with nowhere to hide. We hope to streamline our processes by making Adblock Plus smart enough to recognize intrusive and ‘bad’ ads on its own as well as fighting circumvention.
Facebook (and others) are getting better at disguising ads from ad blockers and using circumvention techniques to show you ads anyway. These tricks disregard your stated preference to block ads! We’re pretty good at coming up with workarounds to these circumvention techniques now, but to respond even faster we’re training an A.I. system to help us.
eyeo, the company that makes Adblock Plus, is developing the Sentinel technology.
Because we thought that it sounds awesome and fearsome. We want aggressive advertisers to know that we’re serious about protecting consumers’ rights. And we want you to know that we have your back.
We use them train Sentinel’s neural networks (it’s a type of Artificial Intelligence). We literally show the Sentinel algorithm what a bunch of ‘bad’ ads look like and tell it “This is an intrusive ad; block anything that looks like this.” This is how the ‘machine learning’ part of AI works.

If you would like your screenshot(s) to be deleted, just type "delete my data" in the Messenger chat at any time (even long after you’ve submitted them).
For now, Sentinel is learning just Facebook ads. But we’ll expand to other ad types really soon.
Because Facebook is by far the most aggressive at disguising ad tags and forcing people to see ads even when they’ve said ‘no’.
Sentinel uses neural networks and machine learning. Neural networks are a type of statistical model (i.e. an algorithm) that is inspired by biological networks in our brains. Along with machine learning and deep learning, neural networks technology is part of the larger ‘Artificial Intelligence’ sector. If you want to go a bit deeper, you can read this overview of ‘ 6 Types of Artificial Neural Networks Being Used In Machine Learning’. Shoot us a note when you’ve finished.
To train our Sentinel neural networks to recognize what ‘bad’ Facebook ads look like, we need lots and lots of examples. So we’re asking our hundreds of millions of users to help out. Be sure to tell your friends to help out also!
We’re experimenting with technologies that will better evade ‘circumvention’ techniques (aka anti-ad blockers like Sourcepoint and Pagefair). These techniques disregard your stated preference to block ads, and find a way to shove ads in front of you anyways. Sentinel is learning how to block ads algorithmically.