For Nour Omar, a student in Cairo with 5,000 Instagram followers, reposting news about the genocide in Gaza is the smallest thing she can do that still feels like something. Usually, her posts draw hundreds of views an hour.

On a Tuesday night in November 2023, a few weeks after an Israeli airstrike on Al-Ahli Hospital in Gaza City killed at least 471 Palestinians, she posted three Instagram Stories in a row: a slide naming the dead from the hospital attack, a 12-second Reel of rooftops sheared open and a hospital donation link tagged #CeasefireNow.

There was an uncanny hush on the feed — just 30 views. Then the Reel was removed for “violating community guidelines.” The link was unopenable, the hashtag unsearchable, the story unavailable to followers. At dawn, an “account status” notification delivered a velvet-gloved threat without explanation: “Some features may be limited.”

She can’t fathom what rule she broke by bearing witness. But she was discovering the platform’s physics of speech, where visibility is less a right than a moving target.

What happened to Nour is now a familiar pattern for Palestinians and their allies online: Peaceful posts about Gaza are quietly buried, flagged or erased. At the same time, dehumanizing content about Palestinians is allowed to circulate freely.

Yet instead of logging off, activists, journalists and ordinary users across the world are piecing together a repertoire of nonviolent digital resistance. Tactics range from how they write to where they post and how they store evidence. Their experiments show that under these conditions, visibility itself becomes something people have to organize for, not something platforms reliably provide.

If social media platforms can code repression into these systems, then resistance has to be coded into how people use them. For Palestinians and their allies, the task is not to wait for a level playing field, but to learn how to keep one another visible inside tools that were never built for their freedom.

Patterned silencing

Digital repression has become a structural filter on Meta’s platforms, deciding who is heard and who is erased. In October and November 2023 alone, Human Rights Watch documented over a thousand cases where Instagram and Facebook removed or suppressed peaceful expressions of solidarity with Palestine.

The patterns were systematic: deleted posts or stories, restricted features, search bans and the quiet throttling of reach known as “shadowbanning.” That same month, Meta’s translation algorithm added the word “terrorist” to Palestinian users’ bios. The company later apologized for the “bug,” but for many, it felt like a slip revealing the machine’s logic.

At the center of this machinery sits Meta, the parent company of Facebook, Instagram and WhatsApp, which functions less like apps and more like global infrastructure. With several billion daily users across its platforms, Meta’s design choices effectively dictate much of the world’s visible reality. When Meta downgrades, deletes or distorts Palestinian content, it is not a marginal glitch on a niche site; it is the main artery of digital communication constricting a people’s ability to speak and be seen.

“These are not random accidents,” said Jalal Abukhater, advocacy manager at 7amleh, the Arab Center for the Advancement of Social Media, a Palestinian digital rights advocacy and research organization that documents online discrimination against Palestinians. “At the heart of these digital rights violations is a blunt truth: Palestinian voices are being systematically silenced. As Israeli forces commit genocide on the ground, platforms mirror that violence in code — deleting their words, their witness, their very traces from the digital world.”

The consequences are immediate and profound: When journalists, activists and human rights defenders lose their social media channels, they lose a layer of protection and the world loses access to real-time, on-the-ground evidence.

That erosion, and its consequences, are vividly documented in 7amleh’s 2024 report, “Erased and Suppressed,” which gathered 20 firsthand testimonies from Palestinian journalists, creators and media outlets describing systematic suppression on Meta platforms. Respondents reported dire ripple effects, including economic loss, psychological stress, career disruption and diminished credibility. When Ehab Al-jariri, radio station 24FM’s editor-in-chief, had his 120,000-follower Facebook account permanently deleted, he lost what he describes as his “paramount protection tool” — the public platform he relied on to push back when he or his newsroom came under attack.

“We feel we are at war with algorithms and their owners,” said Dima Kabaha, an editing assistant at the Palestinian news website Arab48, in the report. “They do not want to see us as human beings.”

Why this keeps happening

These crackdowns don’t recur because moderators are having a series of bad days; they recur because the system is built to produce them.

“Erased and Suppressed” revealed a technical disparity: In the Occupied West Bank and Gaza, Meta’s automated moderation tools needed only an AI confidence threshold of as low as 25 percent to remove content, meaning that Palestinian users’ posts were far more likely to be flagged or taken down.

Leaked internal documents show that Meta also subjects Arabic-language content to far denser and more automated enforcement than content in Hebrew. The company has an Arabic-specific “hostile speech classifier” and fully onboards Arabic content into its precision-tracking systems, while Hebrew has never been brought under the same systematic checks and is handled on a more manual, case-by-case basis. That imbalance means the system misfires in two directions at once: It over-removes legitimate Palestinian speech by treating it as extremist or hateful, and under-removes incitement of violence against Palestinians that circulates in Hebrew. The result is that Palestinians face far higher odds of being punished for speaking than being protected.

The moderation of content about Palestine sits at the intersection of skewed rules, unequal resourcing, and opaque corporate power, so even when one piece is challenged, the underlying machinery remains intact. At the policy level, Meta can appear to course-correct without shifting the ground. In September 2024, its independent oversight board ruled that the standalone phrase “From the river to the sea” should not be automatically treated as hate speech, unless paired with explicit calls for violence. On paper, that signaled a move toward greater protection for political expression. In reality, Palestinian users continued to report that posts supporting Palestinian rights, including uses of “From the river to the sea” that did not praise or promote violence, were still being removed. Policy changed in writing; enforcement, shaped by older assumptions and automated systems, largely did not.

Meta is not an outlier. Amnesty International has warned that across the major networks there is a double pattern at work: a wave of dehumanizing content that advocates violence against Palestinians and at the same time an “over-broad censorship” of Palestinian accounts and pro-Palestinian speech. Several other platforms are known to practice shadowbanning and removals: X is repeatedly named by digital rights groups for allegedly suspending hundreds of Palestinian accounts. YouTube has gone further, recently deleting the channels of three leading Palestinian human rights organizations and thus erasing more than 700 videos documenting alleged Israeli violations. According to Palestinian digital rights group, Sada Social, moves like these strip crucial evidence from public view and punish groups for doing accountability work, rather than for breaking any meaningful “community guideline.”

Tactics of resistance

Forced to work inside an architecture tilted against them, Palestinian activists have become reluctant engineers of their own visibility. They study which words trigger takedowns, anticipate how classifiers read an image, and treat every platform as one node in a wider network rather than a single point of failure. Over time, those experiments have coalesced into a shared playbook that turns everyday posting into a quiet but coordinated form of digital nonviolent resistance. Here are three ways Palestinians and their allies are keeping their stories visible in the face of platform censorship:

1. Recode the message. The first line of defense is not to leave a platform, but to learn how to post in ways the systems cannot fully read. Rather than treating moderation as an absolute wall, Palestinian users and their allies treat it as a language problem. If certain words and formats trigger erasure, then the message has to be rewritten so that people can understand it but machines cannot. What emerges is less a bag of tricks than a shared language, its grammar passed around to signal solidarity and help exhausted users keep posting without vanishing from the feed.

Text is usually the first thing to be recoded. Terms like Palestine, Gaza or Israel are scrambled into symbols and altered spellings, broken across lines or tucked into punctuation so they stay legible to humans but harder for automated systems to parse.

Captions avoid “radioactive” words in the opening line, where scrutiny is highest, and move key information deeper into the text or into comments. Others shift the most sensitive language off the caption entirely and onto the image itself, exploiting the fact that detection remains weaker and less consistent in image-based moderation than in standard text fields.

This is part of a wider phenomenon often called “algospeak”: coded language designed to slip past automated filters while preserving meaning for those who know how to read it.

The visuals are recoded too. Activists front-load carousels with apparently neutral slides (landscapes, quotes, abstract designs) before introducing graphic evidence or explicit political language in later frames, buying enough time for human audiences to see and share before classifiers react.

Symbols also speak. The watermelon, once a sly substitute for a banned flag, now operates as a global shorthand for Palestine solidarity because it bypasses keyword filters and communicates instantly. The symbol’s roots go back to the aftermath of the Six-Day War of 1967, when displaying the Palestinian flag was criminalized in parts of Gaza and the West Bank, and artists turned to the fruit’s red, green, white and black as a proxy. Today, the watermelon image and emoji are that history rendered digital: a compact sign that often slips under automated moderation while still saying exactly what it needs to say.

After her Al-Ahli posts vanished, Nour began to experiment with this kind of coded speech. She stopped putting words like “Gaza” or “ceasefire” in the first line of a caption, broke up Palestine into “P@l3st1ne,” and shifted donation links into replies or made them visible only to close friends. To her followers, the meaning of the posts remained obvious. What changed was the surface of the text, rewritten with the quiet assumption that an unseen machine was always reading over her shoulder.

2. Reroute the message. If the first tactic is to change how you speak, the second is to change where you speak. When one feed goes quiet, activists don’t stop posting; they move the same material through a lattice of channels so that no single platform can make it disappear. A video that risks removal on Instagram is mirrored to X, reposted to Telegram, archived on an independent site and backed up to cloud folders and hard drives. In Gaza and across the Palestinian diaspora, Telegram in particular has become the safety net — a real-time conduit for footage and updates when Meta clamps down.

The logic is simple but powerful: Build redundancy, spread points of failure across apps and time zones, and make sure every peak moment of attention has more than one route to reach those who are ready to see and share.

Rerouting also means leaning into platforms where the balance of visibility looks different. On TikTok, pro-Palestinian content has repeatedly outpaced pro-Israel narratives, not just in isolated moments but over time. A study by Laura Edelson, assistant professor of computer science at Northeastern University and co-director of Cybersecurity for Democracy, a multi-university center for research-driven policy, found that two years after Israel’s war on Gaza began in late 2023, the skew remained stark. In September 2025, for roughly every one pro-Israel post on TikTok, around 17 supported Palestinian perspectives. For activists, that ratio is not a guarantee of safety from censorship, but a signal that some terrains are more open than others, and worth prioritizing when other options close.

But even that advantage now looks precarious. In 2025, under a U.S. law forcing TikTok’s divestment, President Trump approved a deal to transfer the app’s American operations to a consortium of investors, giving the U.S. control of the algorithm. That raises an obvious question: How long can a pro-Palestine skew survive once TikTok’s moderation rules, recommendation system and leadership are restructured under owners with strong political incentives to rein it in?

The rerouting reaches into newer, less tightly policed spaces as well. In virtual worlds like Roblox, Final Fantasy and the Sims, players voice dissent and amplify offline movements. Young users on Roblox have staged digital rallies for Palestine, building protest maps that have been visited hundreds of thousands of times. Inside these worlds, players can march together, unfurl banners and share slogans in an interactive environment that feels less like a comment thread and more like a street. These spaces do not replace traditional organizing, but they extend it, turning games and group chats into parallel routes for solidarity when mainstream channels constrict.

3. Archive the message. If recoding and rerouting are about staying visible in the present, archiving is about refusing to let evidence disappear. Around Gaza, rights groups have urged a shift in mindset from “posting” to “preserving,” reminding users that social media feeds are not archives but fragile display windows controlled by private companies. The stakes of that fragility became clear last month when YouTube deleted three Palestinian organizations’ channels. In an instant, years of painstaking documentation were wiped from one of the world’s largest public platforms, a stark reminder that whoever controls the infrastructure of storage also holds the power over collective memory.

Activists have learned that no matter how widely a post circulates, it is only ever one policy change, or one mass takedown, away from vanishing. So they treat every upload as an opportunity for documentation. Before posts go live, originals, timestamps and screenshots are saved; after takedowns, Stories are exported, links mirrored and copies sent through encrypted channels to volunteer archivists, lawyers and journalists. The goal is not just to keep content online, but to preserve a verifiable chain of custody that can withstand scrutiny in newsrooms, courts or future human rights investigations. Large accounts have adapted accordingly. Pages like @eye.on.palestine, with over 13 million followers, maintain backup profiles and mirror content in anticipation of bans or suspensions.

Breaking up with Meta

For some in the tech world, the most radical act of resistance is no longer learning how to survive inside Meta’s systems, but choosing to leave them. Paul Biggar, software engineer and founder of Tech for Palestine, argues that the problem is too deep to be fixed with smarter tactics alone. After years of watching Palestinian content throttled, flagged and buried, he believes the balance between offense and defense is irreparably skewed.

“Attack is much easier than defense,” Biggar said. “People who are trying to shut us up are much better at it, and will always be much better at it, than the people who are trying to keep their voices heard. You can appeal, you can push back, but during that time, your reach will have gone.”

The result, Biggar said, is that “we have gone from a time where people were able to post information about war crimes … to a point where they have us talking entirely to ourselves,”

Biggar’s conclusion is blunt: The endgame cannot be endless optimization within a hostile platform. It has to be migration and replacement.

“We are not going to be able to prevail while Meta has this place in society,” he said. “So the obvious thing to do is get people onto other platforms … platforms that are designed around speech supporting humanity, speech supporting human rights, where the rules are based on international humanitarian principles.”

He points to models ranging from federated networks like Mastodon, which let communities choose servers with their own rules, to platforms built on emerging open protocols such as Bluesky, to UpScrolled, an Instagram-style platform rooted in humanitarian principles that Tech for Palestine is helping incubate.

In his view, the strategic horizon is not a single “safe” app but an ecosystem: “We need thousands, tens of thousands of pro-Palestine advocates … creating alternate pathways, alternate ecosystems” whose rules are grounded in human rights norms rather than geopolitical alignments. The truly radical move, then, is to stop treating Meta as unavoidable infrastructure, and to ultimately build networks at scale where pro-Palestine speech is not merely tolerated, but protected.

From those learning to slip past Meta’s algorithms to those trying to build entirely new ecosystems, this work is ultimately part of the same project: refusing to let Palestinian testimony be switched off at will. What ultimately gives this struggle its force isn’t a clever workaround in the code — it’s the people who refuse to let each other disappear. Every mirrored link, every forwarded channel, every saved archive is a small act of nonviolent resistance that says: You will be seen.

Here, hope is not a feeling but a practice. It looks like teaching someone how to post without being taken down, backing up footage before it vanishes, shifting audiences into spaces that can’t be switched off with a single policy change. When one route closes, another is opened and the story moves with it. An algorithm can bury a post in the feed, but it cannot bury a network that insists on remembering.

---

ZNetwork is funded solely through the generosity of its readers.

Donate