Internal documents expose Meta’s bias in moderating Arabic and Hebrew content
Occupied Palestine (Quds News Network)- Newly leaked internal documents reveal growing criticism of Meta, the parent company of Facebook, Instagram, and WhatsApp, for its handling of content related to Palestine and Israel. These documents, shared by a former Meta employee according to the Guardian, highlight significant discrepancies in the moderation of Arabic and Hebrew content, suggesting bias in Meta’s content enforcement policies.
According to the internal guidelines, Meta employs a multilayered process for moderating content in various languages, including Arabic and Hebrew. However, the documents indicate that Meta lacks a consistent system to evaluate the accuracy of its Hebrew content moderation, contrasting with the more structured approach for Arabic content. The absence of a “policy precision” measure for Hebrew content means that moderation in the official language of the occupation state of Israel is less systematic, potentially leading to over-enforcement of Arabic content.
The former employee, who requested anonymity due to fears of professional reprisal, criticized Meta’s approach, stating that the company’s hate speech policies related to Palestine are inequitable. This sentiment is echoed by Palestinian advocates and Meta employees alike. A letter signed by over 200 Meta workers states that those raising concerns about the company’s moderation practices face potential retaliation, creating a chilling effect on internal discussions.
The release of these documents comes at a time when Meta and other social platforms are under intense scrutiny for their role in shaping public discourse on the Israeli genocide in Gaza. In June, a coalition of 49 civil society organizations, along with prominent Palestinian voices, stated that Meta enables harmful narratives through its moderation policies. The disparities in language moderation have been a longstanding issue, with Meta previously criticized for disproportionately penalizing Arabic content compared to Hebrew content.
Meta has defended its content moderation practices, claiming that it employs multiple systems to measure enforcement accuracy for Hebrew-language content, including evaluations by Hebrew-speaking reviewers. However, the leaked documents suggest that these measures are not as robust or systematic as those for other languages, raising concerns about potential bias.
As Meta continues to navigate the complexities of moderating content related to the Palestinian cause, these revelations underscore the need for greater transparency and accountability in the company’s content enforcement policies.
