In an update to its hate speech guidelines, Meta will remove content that links the term “Zionist” with antisemitic tropes, calls for harm or denials of existence or other dehumanizing language, the social media giant’s policy forum announced on Tuesday.

In its announcement, Meta said that it “found sharply contrasting views” on how to evaluate the term, noting that the word has “layers of meaning based on its origins and usage today, and may also be highly dependent on context.” 

“For many, the term is a proxy for Jewish people or Israelis,” a statement from the company explained. “This perception is particularly strong when the term is paired with age-old antisemitic tropes – especially those invoking the conspiracy of worldwide Jewish power. Many other stakeholders told us that ‘Zionists’ is a reference to adherents of an ideology, rooted in history, and that today the term is used most often to refer to the Israeli government and its supporters.” 

The company said it “will now remove content that targets ‘Zionists’ with dehumanizing comparisons, calls for harm, or denials of existence on the basis that ‘Zionist’ in those instances often appears to be a proxy for Jewish or Israeli people.”

Several Jewish groups pushed for the policy change, as online antisemitism has climbed 86% in the wake of the Oct. 7 attacks, according to CyberWell, an online antisemitism watchdog. Cyberwell submitted to Meta data-centered analysis on the use of “Zionist” as a replacement for “Jew,” which frequently avoided detection and flagging as antisemitic speech according to the social media platforms’ digital policy and moderation resources. 

Meta and other social media companies have grappled with how to strike a balance between hate speech and free expression, particularly in the aftermath of Oct. 7. The board announced in May it would consider whether Meta “should moderate allegations of criminality based on nationality” after a post relating to the war in Gaza that called all Israelis criminals was removed. The board is asking for the public to offer perspectives on the impact of Meta’s policies on the ability of users to speak out against governmental actions and the effect of posts calling an entire group of people “criminals.” The board is also seeking suggestions for criteria to distinguish between attacks on a group of people and attacks on institutions.

Last week, the social media powerhouse said it will end its blanket ban on the word “shaheed,” an Arabic word commonly translated into English as “martyr.” The lift came after a year-long review by Meta’s Oversight Board, which is funded by Meta but operates independently. The review determined that Meta’s rules on “shaheed” failed to account for the word’s variety of meanings and resulted in the removal of content not aimed at condoning violence.