Facebook saw a huge jump of bots trying to create fake accounts during the first quarter, with the company revealing Thursday that it took down 2.19 billion such accounts over the first three months of this year.

“We’ve seen a steep increase in the creation of abusive, fake accounts on Facebook in the last six months,” the company said as part of its third community standards enforcement report. “We catch most of these accounts within minutes of registration. However, automated attacks have resulted in more of these accounts making it past our initial detection.”

Lazy loaded image
Courtesy of Facebook

Fake accounts taken down by Facebook quarter-over-quarter.

The company took down 99.8% of these accounts proactively, meaning the accounts were detected by automated tools before users had to report them. Accounts that are detected immediately after they’re created aren’t being counted as part of Facebook’s monthly active user base, which was 2.38 billion at the end of March.

The company estimated Thursday that around 5% those users were fake. This means that Facebook ended the quarter with an estimated 119 million fake accounts active on the service. Facebook measures this “prevalence” metric by regularly sampling a subsection of content and accounts.

Asked who might be behind the increased number of fake accounts created on the service, Facebook’s vice president of integrity Guy Rosen said Thursday that the company didn’t have any data to share on the intention of these fake actors. However, he said that most of the activity seems to be driven by commercial activity. “The larger number of fake accounts were driven by spammers,” Rosen said.

Facebook also used its community standards enforcement report to highlight its work on removing posts involving nudity and pornography, hate speech, terrorism-related content and bullying from the service.

The company stressed that it is getting better at proactively removing content that violates its community standards. For instance, 65% of all posts removed for hate speech were caught without users flagging them. A year ago, that number was still at 38%.

For the first time, Facebook also included data on how many posts it restores, either after an internal review, or after a user appeals the take-down. One notable data point: Facebook restored 1.1 million posts taken down for adult nudity and sexual activity in Q1.

During the same quarter, the company removed 19.4 million pieces of content for the same reasons. However, Facebook cautioned that some of the content restored in Q1 may have been taken down during the previous quarter.

Facebook noted that the amount of child pornography taken down by the company went down during the quarter, but also admitted that it may not have been fully prepared to catch every instance.

“A bug impacted our ability to store hashes of violating videos that were already removed, making it harder to detect other instances of the same video if it was shared,” the company wrote in its report. “We fixed this bug and are working to remove anything we may have missed.”