Featured Article

Meta has a moderation bias problem, not just a ‘bug,’ that’s suppressing Palestinian voices

Instagram users suspect Meta of shadowbanning content about Palestine. The platform has a history of inherent bias.

Comment

Instagram's logo with eyes and arms in a shrugging position.
Image Credits: Bryce Durbin / TechCrunch

Earlier this year, Palestinian-American filmmaker Khitam Jabr posted a handful of Reels about her family’s trip to the West Bank. In the short travel vlogs, Jabr shared snippets of Palestinian culture, from eating decadent meals to dancing at her niece’s wedding. 

“I hadn’t been in a decade, so it’s just like, life abroad,” Jabr told TechCrunch.

“I would get [anti-Palestine] comments,” she recalled. “And I couldn’t respond [to them] or use my account for 24 hours. I wasn’t even posting anything about the occupation. But fast forward to now and the same shit’s happening.” 

In the aftermath of Hamas’ attack on Israelis, Israel’s retaliatory airstrikes and total blockade — cutting access to electricity, water and vital supplies — have devastated Gaza. In response to the escalating violence, Meta said that it is closely monitoring its platforms for violations and may inadvertently flag certain content, but it never intends to “suppress a particular community or point of view.” Content praising or supporting Hamas, which governs Gaza and is designated as a terrorist organization by the United States and the European Union, is expressly forbidden on Meta’s platforms. 

As the humanitarian crisis in Gaza grows more dire, many social media users suspect Instagram of censoring content about the besieged Palestinian territory, even if that content doesn’t support Hamas. Users have also complained that they’ve been harassed and reported for posting content about Palestine, regardless of whether or not it violates Meta’s policies. Jabr, for example, suspects that Instagram restricted her for 24 hours because other users reported her Palestine travel videos. Most recently, Instagram users accused Meta of “shadowbanning” their Stories about Palestine. 

https://www.tiktok.com/@khitam.mp4/video/7288812363568925995

It’s the latest in a lengthy history of incidents on Meta platforms that reflect an inherent bias against Palestinian users in its processes, as documented by years of complaints from both inside and outside the company. The company may not intentionally suppress specific communities, but its moderation practices often disproportionately affect Palestinian users. 

For instance, Meta struggles to navigate the cultural and linguistic nuances of Arabic, a language with over 25 dialects, and has been criticized for neglecting to adequately diversify its language resources. The company’s black-and-white policies often preclude it from effectively moderating any nuanced topic, like content that discusses violence without condoning it. Advocacy groups have also raised concerns that Meta’s partnerships with government agencies, such as the Israeli Cyber Unit, politically influence the platform’s policy decisions. 

During the last violent outbreak between Hamas and Israel in 2021, a report commissioned by Meta and conducted by a third party concluded that the company’s actions had an “adverse human rights impact” on Palestinian users’ right to freedom of expression and political participation.

Meta communications director Andy Stone declined to comment on the record, and pointed TechCrunch to Meta’s newsroom post about the conflict, which was updated on Oct. 18 with a statement addressing the shadowbanning claims.

“Our policies are designed to keep people safe on our apps while giving everyone a voice,” the post stated. “We apply these policies equally around the world and there is no truth to the suggestion that we are deliberately suppressing voice … We can make errors and that is why there is an appeals process for people to tell us when they think we have made the wrong decision, so we can look into it.”

The belief that Meta shadowbans, or limits the visibility of, content about Palestine is not new. In an Instagram Story last year, supermodel and activist Bella Hadid, who is of Palestinian descent, alleged that Instagram “disabled” her from posting content on her Story “pretty much only when it is Palestine based.” She said she gets “immediately shadowbanned” when she posts about Palestine, and her Story views drop by “almost 1 million.” 

Meta blamed technical errors for the removal of posts about Palestine during the 2021 conflict. When reached for comment about these recent claims of shadowbanning, a representative for the company pointed TechCrunch to a Threads post by Stone. 

“We identified a bug impacting all Stories that re-shared Reels and Feed posts, meaning they weren’t showing up properly in people’s Stories tray, leading to significantly reduced reach,” Stone said. “This bug affected accounts equally around the globe and had nothing to do with the subject matter of the content — and we fixed it as quickly as possible.” 

But many are frustrated that Meta continues to disproportionately suppress Palestinian voices. Leen Al Saadi, a Palestinian journalist currently based in Jordan and host of the podcast “Preserving Palestine,” said she is used to “constantly being censored.” Her Instagram account was restricted last year after she posted a trailer for the podcast’s first episode, which discussed a documentary about Palestinian street art under occupation. 

“Palestinians are currently undergoing two wars,” Al Saadi said. “The first is with their legal occupier. The second war is with the entire Western media landscape, and when I say the entire landscape, I mean social media.” 

Meta’s alleged shadowbanning

Instagram users accuse Meta of suppressing more than just Stories related to Palestine. 

Creators say engagement on their posts tanked specifically after they publicly condemned Israel’s response to the Hamas attack as excessively violent. Some, like Jabr, say they were restricted from posting or going live, while others say Instagram flagged their content as “sensitive,” limiting its reach. Users also allege their posts were flagged as “inappropriate” and removed, even if the content adhered to Instagram’s Community Guidelines

Meta’s representative didn’t address the other accusations of censorship beyond just Story visibility and did not respond to TechCrunch’s follow-up questions. When asked for comment, Stone directed TechCrunch to Meta’s updated newsroom post, which said Meta identified and fixed a “bug” that prevented users from going live.

“This was also a global issue that was fixed within a few hours. We understand people rely on these tools and we’re sorry to anyone who felt the impact of these issues,” the post stated. 

It’s unclear if this “bug” impacted accounts posting content unrelated to Gaza. Instagram users have posted screenshots showing that Stories about Palestine have received significantly fewer views than other Stories posted on the same day, and allege that their view counts went back up when they posted content unrelated to the conflict. 

https://twitter.com/arxbprince/status/1713657399976866089

A user based in Egypt, who asked to stay anonymous for fear of harassment, said her posts usually get around 300 views, but when she started posting pro-Palestine content after the Hamas attack earlier this month, her stories would only get one to two views. 

“It happened to all my friends, too,” she continued. “Then we noticed that posting a random pic would get higher views. So by posting a random pic, then a pro-Palestine post, would increase the views.” 

Another Instagram user based in the United Kingdom, who also asked to stay anonymous out of fear of harassment, said that his view count returned to normal when he posted a cat photo. 

“My stories went from 100s of views to zero or a handful,” he said. “I’ve had to post intermittent non-Gaza content in order to ‘release’ my stories to be viewed again.” 

It isn’t just Stories. The Arab Center for Social Media Advancement (7amleh), which documents cases of Palestinian digital rights violations and works directly with social media companies to appeal violations, told TechCrunch it has received reports of Instagram inconsistently filtering comments containing the Palestinian flag emoji. Users report that Instagram has flagged comments containing the emoji as “potentially offensive,” hiding the comment.

Meta initially did not respond to follow-up requests for comment. When asked for comment, Stone pointed TechCrunch to Meta’s updated newsroom post, which did not directly address issues related to comments containing the Palestinian flag. 

The organization has also received countless reports of Meta flagging and restricting Arabic content, even if it’s posted by news outlets. Jalal Abukhater, 7amleh’s advocacy manager, said that the organization has documented multiple cases of journalists on Instagram reporting the same news in Arabic, Hebrew and English, but only getting flagged for their Arabic content. 

“It’s literally journalistic content, but the same wording in Hebrew and English does not get restricted,” Abukhater said. “As if there’s better moderation for those languages, and more careless moderation for Arabic content.” 

Stone declined to comment on allegations of Meta disproportionately flagging Arabic news content, instead directing TechCrunch to the updated newsroom post — which did not directly address the allegations.

And as the Intercept reported, Instagram and Facebook are flagging images of the al-Ahli Hospital, claiming that the content violates Meta’s Community Guidelines on nudity or sexual activity.

The Community Guidelines are enforced inconsistently, particularly when it comes to content related to Palestine. Al Saadi recently tried to report a comment that said she should be “raped” and “burned alive” — left in response to her comment on a CNN post about the conflict — but in screenshots reviewed by TechCrunch, Instagram said that it didn’t violate the platform’s Community Guidelines against violence or dangerous organizations. 

“The restrictions on content, especially the content that relates to Palestine, is heavily politicized,” Abukhater said. “It feeds into the bias against Palestinian narrative genuinely. It really takes the balance against Palestinians in a situation where there’s a huge asymmetry of power.”

A history of suppression

Content about Palestine is disproportionately scrutinized, as demonstrated during the last severe violent outbreak between Hamas and Israel two years ago. Amid the violence following the May 2021 court ruling to evict Palestinian families from Sheikh Jarrah, a neighborhood in occupied East Jerusalem, users across Facebook and Instagram accused Meta of taking down posts and suspending accounts that voiced support for Palestinians. 

The digital rights nonprofit Electronic Frontier Foundation (EFF) described Meta’s actions in 2021 as “systemic censorship of Palestinian voices.” In its 2022 report of Palestinian digital rights, 7amleh said that Meta is “still the most restricting company” compared to other social media giants in the extent of its moderation of the Palestinian digital space. 

https://www.instagram.com/p/CyWCzLGs3dg/?utm_source=ig_web_copy_link&igshid=MzRlODBiNWFlZA==

Meta forbids support of terrorist organizations, like most social media companies based in the U.S., but struggles to moderate content around it, from user discourse to journalistic updates. This policy, along with the company’s partnership with Israel to monitor posts that incite violence, complicates things for Palestinians living under Hamas’ governance. As EFF points out, something as simple as Hamas’ flag in the background of an image can result in a strike. 

Jillian York, the director for international freedom of expression for EFF, blames automation and decisions made by “minimally trained humans” for the inconsistency. Meta’s zero tolerance policy and imprecise enforcement often suppress content from or about conflict zones, she said. The site’s moderation issues have negatively affected multiple non-English speaking regions, including Libya, Syria and Ukraine. 

“These rules can prevent people from sharing documentation of human rights violations, documentation of war crimes, even just news about what’s happening on the ground,” York continued. “And so I think that is what is the most problematic right now about that particular rule, and the way that it’s enforced.” 

Over the 13 days leading up to the ceasefire between Hamas and Israel, 7amleh documented more than 500 reports of Palestinian “digital rights violations,” including the removal and restriction of content, hashtags and accounts related to the conflict. 

Meta blamed some of the instances of perceived censorship to technical issues, like one that prevented users in Palestine and Colombia from posting Instagram Stories. It attributed others to human error, like blocking the hashtag for Al-Aqsa Mosque, the holy site where Israeli police clashed with Ramadan worshippers, because it was mistaken for a terrorist organization. The company also blocked journalists in Gaza from WhatsApp without explanation. 

The same month, a group of Facebook employees filed internal complaints accusing the company of bias against Arab and Muslim users. In internal posts obtained by BuzzFeed News, an employee attributed the bias to “years and years of implementing policies that just don’t scale globally.” 

At the recommendation of its Oversight Board, Meta conducted a third-party due diligence report about the platform’s moderation during the May 2021 conflict. The report found that Arabic content was flagged as potentially violating at significantly higher rates than Hebrew content was, and was more likely to be erroneously removed. The report noted that Meta’s moderation system may not be as precise for Arabic content as it was for Hebrew content, because the latter is a “more standardized language,” and suggested that reviewers may lack the linguistic and cultural competence to understand less common Arabic dialects like Palestinian Arabic. 

Has anything improved?

Meta committed to implementing policy changes based on the report’s recommendations, such as updating its keywords associated with dangerous organizations, disclosing government requests to remove content and launching a hostile speech classifier for Hebrew content. Abukhater added that Meta has improved its response to harassment, at least in comparison to other social media platforms like X (formerly Twitter). Although harassment and abuse are still rampant on Instagram and Facebook, he said, the company has been responsive to suspending accounts with patterns of targeting other users. 

The company has also made more contact with regional Palestinian organizations since 2021, York added, but it’s been slow to implement recommendations from EFF and other advocacy groups. It’s “very clear” that Meta is not putting the same resources behind Arabic and other non-English languages, York said, compared to the attention Meta gives to countries that have the most regulatory pressure. Moderation of English and other European languages tends to be more comprehensive, for example, because the EU enforces the Digital Services Act

EU also warns Meta over illegal content, disinfo targeting Israel-Hamas war

In Meta’s response to the report, Miranda Sissons, the company’s director of human rights, said that Meta was “assessing the feasibility” of reviewing Arabic content by dialect. Sissons said that the company has “large and diverse teams” who understand “local cultural context across the region,” including in Palestine. Responding to the escalating violence earlier this month, Meta stated that it established a “special operations center” staffed with fluent Hebrew and Arabic speakers to closely monitor and respond to violating content. 

Despite Meta’s apparent efforts to diversify its language resources, Arabic is still disproportionately flagged as violating — like in the case of journalists reporting news in multiple languages. 

“The balance of power is very fixed, in reality, between Israelis and Palestinians,” Abukhater said. “And this is something that today is reflected heavily on platforms like Meta, even though they have human rights teams releasing reports and trying to improve upon their policies. Whenever an escalation like the one we’re experiencing now happens, things just go back to zero.”

https://www.instagram.com/p/Cyblkb0sQZm

And at times, Meta’s Arabic translations are completely inaccurate. This week, multiple Instagram users raised concerns over the platform mistranslating the relatively common Arabic phrase “Alhamdulillah,” or “Praise be to God.” In screen recordings posted online, users found that if they included “Palestinian” and the corresponding flag emoji in their Instagram bio along with the Arabic phrase, Instagram automatically translated their bio to “Palestinian terrorists – Praise be to Allah” or “Praise be to God, Palestinian terrorists are fighting for their freedom.” When users removed “Palestinian” and the flag emoji, Instagram translated the Arabic phrase to “Thank God.” Instagram users complained that the offensive mistranslation was active for hours before Meta appeared to correct it.

Shayaan Khan, a TikTok creator who posted a viral video about the mistranslation, told TechCrunch that Meta’s lack of cultural competence isn’t just offensive, it’s dangerous. He said that the “glitch” can fuel Islamophobic and racist rhetoric, which has already been exacerbated by the war in Gaza. Khan pointed to the fatal stabbing of Wadea Al-Fayoume, a Palestinian-American child whose death is being investigated as a hate crime

Meta did not respond to TechCrunch’s request for comment about the mistranslation. Abukhater said that Meta told 7amleh that a “bug” caused the mistranslation. In a statement to 404 Media, a Meta spokesperson said that the issue had been fixed. 

“We fixed a problem that briefly caused inappropriate Arabic translations in some of our products,” the statement said, “We sincerely apologize that this happened.”

@ytkingkhan

Meta definitely needs to address this (though I couldnt find an official TikTok account for them) #palestine #arab #desi #muslim

♬ original sound – Khan Man

As the war continues, social media users have tried to find ways around the alleged shadowbanning on Instagram. Supposed loopholes include misspelling certain words, like “p@lestine” instead of “Palestine,” in hopes of bypassing any content filters. Users also share information about Gaza in text superimposed over unrelated images, like a cat photo, so it won’t be flagged as graphic or violent content. Creators have tried to include an emoji of the Israeli flag or tag their posts and Stories with #istandwithisrael, even if they don’t support the Israeli government, in hopes of gaming engagement. 

Al Saadi said that her frustration with Meta is common among Palestinians, both in occupied territories and across the diaspora. 

“All we’re asking for is to give us the exact same rights,” she said. “We’re not asking for more. We’re literally just asking Meta, Instagram, every single broadcast channel, every single media outlet, to just give us the respect that we deserve.” 

Dominic-Madori Davis contributed to this story’s reporting.

EU fires urgent warning at Elon Musk’s X over illegal content and disinformation following Hamas attacks

Hacktivism erupts in response to Hamas-Israel war

Meta says it’s prioritizing livestreaming checks during Israel-Hamas war

More TechCrunch

Featured Article

CIOs’ concerns over generative AI echo those of the early days of cloud computing

CIOs trying to govern generative AI have the same concerns they had about cloud computing 15 years ago, but they’ve learned some things along the way.

2 hours ago
CIOs’ concerns over generative AI echo those of the early days of cloud computing

It sounds like the latest dispute between Apple and Fortnite-maker Epic Games isn’t over. Epic has been fighting Apple for years over the company’s revenue-sharing requirements in the App Store.…

Epic Games CEO promises to ‘fight’ Apple over ‘absurd’ changes

As deep-pocketed companies like Amazon, Google and Walmart invest in and experiment with drone delivery, a phenomenon reflective of this modern era has emerged. Drones, carrying snacks and other sundries,…

What happens if you shoot down a delivery drone?

A police officer pulled over a self-driving Waymo vehicle in Phoenix after it ran a red light and pulled into a lane of oncoming traffic, according to dispatch records. The…

Waymo robotaxi pulled over by Phoenix police after driving into the wrong lane

Welcome back to TechCrunch’s Week in Review — TechCrunch’s newsletter recapping the week’s biggest news. Want it in your inbox every Saturday? Sign up here. This week, Figma CEO Dylan…

Figma pauses its new AI feature after Apple controversy

We’ve created this guide to help parents navigate the controls offered by popular social media companies.

How to set up parental controls on Facebook, Snapchat, TikTok and more popular sites

Featured Article

You could learn a lot from a CIO with a $17B IT budget

Lori Beer’s work is a case study for every CIO out there, most of whom will never come close to JP Morgan Chase’s scale, but who can still learn from how it goes about its business.

24 hours ago
You could learn a lot from a CIO with a $17B IT budget

For the first time, Chinese government workers will be able to purchase Tesla’s Model Y for official use. Specifically, officials in eastern China’s Jiangsu province included the Model Y in…

Tesla makes it onto Chinese government purchase list

Generative AI models don’t process text the same way humans do. Understanding their “token”-based internal environments may help explain some of their strange behaviors — and stubborn limitations. Most models,…

Tokens are a big reason today’s generative AI falls short

After multiple rejections, Apple has approved Fortnite maker Epic Games’ third-party app marketplace for launch in the EU. As now permitted by the EU’s Digital Markets Act (DMA), Epic announced…

Apple approves Epic Games’ marketplace app after initial rejections

There’s no need to worry that your secret ChatGPT conversations were obtained in a recently reported breach of OpenAI’s systems. The hack itself, while troubling, appears to have been superficial…

OpenAI breach is a reminder that AI companies are treasure troves for hackers

Welcome to Startups Weekly — TechCrunch’s weekly recap of everything you can’t miss from the world of startups. Sign up here to get it in your inbox every Friday. Most…

Space for newcomers, biotech going mainstream, and more

Elon Musk’s X is exploring more ways to integrate xAI’s Grok into the social networking app. According to a series of recent discoveries, X is developing new features like the…

X plans to more deeply integrate Grok’s AI, app researcher finds

We’re about four months away from TechCrunch Disrupt 2024, taking place October 28 to 30 in San Francisco! We could not bring you this world-class event without our world-class partners…

Meet Brex, Google Cloud, Aerospace and more at Disrupt 2024

In its latest step targeting a major marketplace, the European Commission sent Amazon another request for information (RFI) Friday in relation to its compliance under the bloc’s rulebook for digital…

Amazon faces more EU scrutiny over recommender algorithms and ads transparency

Quantum Rise, a Chicago-based startup that does AI-driven automation for companies like dunnhumby (a retail analytics platform for the grocery industry), has raised a $15 million seed round from Erie…

Quantum Rise grabs $15M seed for its AI-driven ‘Consulting 2.0’ startup

On July 4, YouTube released an updated eraser tool for creators so they can easily remove any copyrighted music from their videos without affecting any other audio such as dialog…

YouTube’s updated eraser tool removes copyrighted music without impacting other audio

Airtel, India’s second-largest telecom operator, on Friday denied any breach of its systems following reports of an alleged security lapse that has caused concern among its customers. The telecom group,…

India’s Airtel dismisses data breach reports amid customer concerns

According to a recent Dealroom report on the Spanish tech ecosystem, the combined enterprise value of Spanish startups surpassed €100 billion in 2023. In the latest confirmation of this upward trend, Madrid-based…

Spain’s exposure to climate change helps Madrid-based VC Seaya close €300M climate tech fund

Forestay, an emerging VC based out of Geneva, Switzerland, has been busy. This week it closed its second fund, Forestay Capital II, at a hard cap of $220 million. The…

Forestay, Europe’s newest $220M growth-stage VC fund, will focus on AI

Threads, Meta’s alternative to Twitter, just celebrated its first birthday. After launching on July 5 last year, the social network has reached 175 million monthly active users — that’s a…

A year later, what Threads could learn from other social networks

J2 Ventures, a firm led mostly by U.S. military veterans, announced on Thursday that it has raised a $150 million second fund. The Boston-based firm invests in startups whose products…

J2 Ventures, focused on military healthcare, grabs $150M for its second fund

HealthEquity said in an 8-K filing with the SEC that it detected “anomalous behavior by a personal use device belonging to a business partner.”

HealthEquity says data breach is an ‘isolated incident’

Roll20 said that on June 29 it had detected that a “bad actor” gained access to an account on the company’s administrative website for one hour.

Roll20, an online tabletop role-playing game platform, discloses data breach

Fisker has a willing buyer for its remaining inventory of all-electric Ocean SUVs, and has asked the Delaware Bankruptcy Court judge overseeing its Chapter 11 case to approve the sale.…

Fisker asks bankruptcy court to sell its EVs at average of $14,000 each

Teddy Solomon just moved to a new house in Palo Alto, so he turned to the Stanford community on Fizz to furnish his room. “Every time I show up to…

Fizz, the anonymous Gen Z social app, adds a marketplace for college students

With increasing competition for what is, essentially, still a small number of hard tech and deep tech deals, Sidney Scott realized it would be a challenge for smaller funds like…

Why deep tech VC Driving Forces is shutting down

A guide to turn off reactions on your iPhone and Mac so you don’t get surprised by effects during work video calls.

How to turn off those silly video call reactions on iPhone and Mac

Amazon has decided to discontinue its Astro for Business device, a security robot for small- and medium-sized businesses, just seven months after launch.  In an email sent to customers and…

Amazon retires its Astro for Business security robot after only 7 months

Hiya, folks, and welcome to TechCrunch’s regular AI newsletter. This week in AI, the U.S. Supreme Court struck down “Chevron deference,” a 40-year-old ruling on federal agencies’ power that required…

This Week in AI: With Chevron’s demise, AI regulation seems dead in the water