Artificial IntelligenceCryptocurrencyMetaverseCybersecurityTech Forward

Here’s How Twitter Is Trying to Stop Bullying and Abuse on Its Service

February 7, 2017, 9:36 PM UTC
Twitter Illustrations Ahead Of Earnings Release
The Twitter logo behind an Apple iPhone 6s displaying the Twitter app.
Michael Nagle—Bloomberg/Getty Images

It might become more difficult for bad actors to hurl abusive comments at random people on Twitter.

Twitter said Tuesday that it made three new changes to its social messaging service that would presumably help its users avoid miscreants who clog their Twitter feeds with hurtful messages.

Critics have long chastised Twitter (TWTR) for failing to prevent abuse on the popular messaging service. In July, Twitter suspended the accounts of people who harassed actress Leslie Jones after the Ghostbusters actress publicly announced she would leave Twitter because of the abuse.

Even Twitter’s former CEO Dick Costolo recently said that he wished he could have done more to stop online abuse during his tenure, according to news website Axios.

“I take responsibility for not taking the bull by the horns,” Costolo said, according to the report.

For more about Twitter, watch:

With the new changes, people who have been permanently suspended for previous bad behavior will be prevented from creating new accounts, wrote Twitter vice president of engineering Ed Ho in a blog post. Ho added that Twitter is “taking steps to identify” these troublemakers, but he did not elaborate on how exactly Twitter would do so.

When contacted by Fortune, Twitter said that it did not want specifically detail how it plans to identify troublemakers because it doesn’t want those people to circumvent its methods.

Twitter is also developing a so-called “safe search” feature, which prevents users from seeing tweets with “potentially sensitive content” as well as tweets from blocked accounts or accounts users have muted from their individual feeds.

Ho also noted that Twitter is attempting to make “potentially abusive and low-quality replies” appear much lower in Twitter conversations than they would have been in the past. Regarding what constitutes low-quality replies, Twitter told Fortune that these kinds of replies could include confrontational tweets, sent from a new account without many followers, that are aimed at a person who doesn’t follow the newly created account.

Twitter said this is a big task considering how many tweets the service receives, but that the use of machine learning technology should help the company get better at identifying problematic tweets.

Twitter CEO Jack Dorsey said in tweet the new features were “a completely new approach to abuse on Twitter.”

As of now, Twitter isn’t detailing how exactly these new changes will be implemented. Ho said that some “some changes will be visible and some less so” to users.

Get Data Sheet, Fortune’s technology newsletter.

Fortune contacted Twitter for more information and will update this story if it responds.

Story updated at 2:00 PM PST with Twitter comments.