Pretty much ever since Instagram started booming in popularity, users have turned to automated apps that generate likes and follows from fake users to make it seem like their posts are gaining more attention than they actually are. Now, Instagram is trying to crack down on these fishy apps using machine learning.

In a blog post, Instagram said that starting today it will begin removing activity generated “from accounts that use third-party apps to boost their popularity.” That includes likes, follows, and comments generated by these accounts, as they go against Instagram’s Terms of Use. TechCrunch names Archie, InstarocketProx, and Boostio as services that do this — and charge anywhere from $10 to $45 per month to do this.

Instagram will spot this type of activity through new machine learning models it’s built, though the post doesn’t give any specifics on how they were built and what signs of fake activity they’ll take into account.

The user will receive a notification from Instagram that the fake activity in question has been removed, and will be asked to change their password, as some of these apps ask for users’ names and passwords.

Instagram also announced in August that it was taking aim at another type of inauthentic activity — accounts that purport to be run by someone in a country other than the one they’re actually located in — with the addition of an “about this account” tab.

At the time, Instagram was dealing with an issue where a group of users reported that their accounts were hacked by users with a Russian email address. The issue also prompted Instagram to add support for third-party two-factor authentication apps.

The battle to rid Instagram of both inauthentic activity and accounts is still far from over though, as Instagram repeatedly had to remove groups of fake accounts linked to Russian and Iranian trolls in the months and weeks leading up to the U.S. midterm elections.