olivewire.org

Social Media Surveillance and the Digital Rights of Palestinians

On a morning in late 2017, Israeli police arrived at the home of a Palestinian man in the West Bank — not because of anything he had done, but because of something he had written. His Facebook post, which contained the Arabic phrase yudbahukum — loosely, “they will slaughter you” — had been flagged by an algorithmic policing system. The post was later confirmed to be a mistranslation. He had typed “good morning,” a common poetic Arabic greeting. He was arrested, held, and released. No charges were filed. The algorithm had been wrong, and he had paid for the error with his freedom, however briefly.

That case, widely reported at the time and cited by digital rights researchers, became a shorthand for something larger: the way Palestinian speech, on platforms designed in California and policed through Israeli military and civilian intelligence frameworks, is processed through systems that treat Arabic as inherently suspicious and Palestinian expression as a potential security threat.

Predictive Policing and the Architecture of Digital Arrest

Israel’s use of social media data to anticipate and prevent what it terms “lone wolf” attacks accelerated significantly after 2015, during the period known as the “knife intifada.” The Israeli Police and the Shin Bet began deploying algorithmic tools — most prominently a system called Blue Wolf and a social media monitoring platform known as Voyager Analytics — to scan Palestinian Facebook and Twitter accounts for keywords, phrases, and behavioral patterns that might predict violence.

The legal architecture enabling these arrests is the Emergency Defense Regulations of 1945, inherited from British Mandate law, alongside provisions of Israeli Military Order 101, which criminalizes political expression, incitement, and “hostile propaganda” in the occupied territories. Under these frameworks, a Facebook post can constitute a prosecutable offense. According to documentation compiled by Addameer Prisoner Support and Human Rights Association, hundreds of Palestinians have been arrested in connection with social media posts since 2015, with many held in administrative detention — imprisonment without charge or trial — for months.

The “good morning” mistranslation case is not an isolated glitch. It points to a structural problem: the natural language processing models underpinning these surveillance tools were largely trained on Hebrew and English data sets. Arabic — particularly Palestinian colloquial Arabic, with its regional idioms, metaphors, and poetic registers — is systematically misread. The consequences of those misreadings are borne entirely by Palestinians.

Meta’s Own Audit: Findings the Company Would Rather Not Lead With

In 2021, Meta (then still operating publicly as Facebook) commissioned the consultancy Business for Social Responsibility (BSR) to conduct an independent human rights impact assessment of its policies and their effect on Palestinian users and civil society. The resulting report, published in May 2021 during the Israeli military assault on Gaza and the crackdown on protests in Jerusalem, was notable for what it acknowledged.

BSR found that Meta’s content moderation systems had engaged in the over-removal of Palestinian content, including posts documenting Israeli military actions, footage of Sheikh Jarrah eviction proceedings, and coverage of protests at Al-Aqsa Mosque. The report documented errors across Arabic-language moderation, concluding that Meta’s automated systems and human reviewers both demonstrated patterns that disproportionately suppressed Palestinian political speech while leaving up content targeting Palestinians.

BSR also found that Meta had cooperated with Israeli government takedown requests at a notably high rate. The Israeli government, according to data Meta itself publishes in its transparency reports, consistently ranks among the top requestors of content removal globally. BSR’s assessment noted that this cooperation, combined with flawed automated moderation, created a compounding effect: Palestinian voices were removed while the content surveilling, targeting, or inciting against them often remained.

Meta acknowledged the report’s findings and committed to a series of remediation steps, including improving Arabic-language moderation and creating a dedicated escalation channel for Palestinian civil society. Researchers and digital rights organizations have remained skeptical about the pace and depth of implementation.

7amleh and Sada Social: Documenting What Platforms Won’t

Into the gap left by platform opacity and state secrecy, Palestinian civil society has built its own monitoring infrastructure. Two organizations in particular have produced the most systematic documentation of the problem.

7amleh — the Arab Center for the Advancement of Social Media, based in Haifa, publishes annual index reports tracking violations of Palestinian digital rights. Its Index of Israeli Violations of Palestinian Digital Rights catalogs content removal incidents, account suspensions, targeted hacking attempts, and surveillance operations. In its 2022 report, 7amleh documented over 450 cases of Palestinian content being suppressed or removed across major platforms, with Facebook and Instagram accounting for the majority of incidents. The report also noted a pattern of Palestinian journalists and human rights defenders having their accounts suspended during periods of heightened military activity — precisely when documentation of events on the ground is most critical.

Sada Social, a Ramallah-based digital monitor established to track violations of Palestinian digital rights in real time, operates a rapid-response documentation system. During the May 2021 escalation, Sada Social recorded dozens of incidents of live-streamed content from Sheikh Jarrah being taken down within minutes of posting, often before the footage could be archived or shared. The organization flagged these cases directly to Meta and filed formal complaints — a process that, as Sada Social’s coordinators have noted publicly, is slow and frequently yields no substantive response.

Together, 7amleh and Sada Social have created a record that challenges the framing, often advanced by platforms and by Israeli authorities alike, that content moderation is a neutral, technical process. The data shows it is neither.

The Stakes: Speech, Visibility, and the Right to Document

For Palestinians under occupation, the stakes of digital surveillance extend beyond the inconvenience of a deleted post. Social media is, in many contexts, the primary mechanism through which human rights violations are documented, witnesses reach journalists, families learn about arrests, and political organizing occurs. When those channels are surveilled by an occupying power and simultaneously over-moderated by platforms deferential to that power’s government requests, the effect is a compounding silence.

Legal scholars including Noura Erakat have argued that digital suppression of Palestinian expression must be understood within the framework of international humanitarian law — specifically, the obligations of an occupying power to protect civilian populations, which extend to the conditions that enable civilian political life. The International Federation of Journalists and Reporters Without Borders have both cited Israeli surveillance and platform over-moderation as threats to press freedom in the occupied territories.

The “good morning” case is years old now. The man was released. The algorithm, as far as researchers can determine, was never corrected. And every day, across the West Bank, Gaza, and the Palestinian communities inside Israel, people choose their words on Facebook with the knowledge — or the suspicion — that someone, human or machine, is reading them in ways they were never meant to be read.

Sources

Étiqueté , , , , ,

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *