It's not a glitch: how Meta systematically censors Palestinian voices

This Access Now report documents patterns of content moderation against Palestinian and pro-Palestinian voices on Facebook and Instagram following October 7, 2023. Drawing on Human Rights Watch's tally of 1,049 content removals across more than 60 countries and the Palestinian Observatory for Digital Rights Violations' log of roughly 1,043 incidents between October 2023 and February 2024, plus direct testimony from affected users, the report identifies four recurring patterns: arbitrary content removal, suspension of prominent accounts, restrictions on pro-Palestinian users, and shadow-banning.

The report ties these outcomes to Meta's "Dangerous Individuals and Organizations" policy and to inconsistent automated enforcement, citing earlier leaked findings that Arabic-language moderation systems had a high error rate and that detection thresholds for Palestine-related content were lowered well below those for other Middle Eastern content. It also points to approved advertisements containing violent rhetoric against Palestinians as evidence of uneven enforcement between Arabic and Hebrew content.

Access Now concludes that the pattern reflects systemic and discriminatory moderation failures rather than isolated technical errors, and calls on Meta to overhaul its policies and enforcement practices.