“It’s clear that these concepts are deeply linked to organized hate groups and have no place on our services,” Facebook said in a blog post, adding that white nationalism and white separatism weren’t targeted before “because we were thinking about broader concepts of nationalism and separatism—things like American pride and Basque separatism, which are an important part of people’s identity.”

After speaking with experts on civility and race relations—those who have condemned Facebook for accepting white nationalism and white separatism, even after the deadly white supremacy movement in Charlottesville—Facebook said it determined these phrases “cannot be meaningfully separated from white supremacy and organized hate groups.”

Individuals who search for terms related to white supremacy on Facebook or Instagram will be directed to Life After Hate, a group founded by former extremists that aim to help people leave hate groups. The new regulations will go into effect next week, said Facebook.

According to Motherboard, which first reported the change Wednesday, “implicit and coded” white nationalism and white separatism will not be immediately banned, partly because it’s more difficult to detect. Facebook’s artificial intelligence tools only detect and remove 50% of hate speech moments after it’s posted, said the Washington Post, meaning such content could go viral before moderators are able to take it down.

While civil rights groups are praising Facebook’s new regulation, many say that the social media giant has been too slow to act. For example, white supremacists organized on Facebook, other social networks and private chat rooms ahead of the Charlottesville rally in 2017, and while Facebook attempted to tighten regulations afterward, hate still lives on the platform.

A recent hate-fueled mass shooting in New Zealand that killed 50 people was live-streamed on Facebook, prompting further criticism of the company.