Algorithms of Discrimination In Algeria

Djalfa City Becomes the Target of Electoral Mockery

A wave of sarcasm and bullying targeting Djelfa province spread widely on social media. The province has been the subject of mockery before and after every election. An Algerian activist posted a sarcastic video on TikTok, set to the soundtrack of a historical film, to share his thoughts on the most recent presidential elections, held on September 7, 2024. In the video, the activist announced that he had not participated in the elections, adding “sarcastically”: "My hands are clean. Did you think I lived in Djelfa?”, mockingly referring to the fact that the residents of Djelfa would have their hands marked with the “blue ink” used in elections.

The video sarcastically referenced a scene from the film Dawn of Islam, in which a main character pretends his hand was paralysed after destroying a statue of the gods and incurring “God's” wrath. The video reflects sarcastic jabs directed at the residents of Djelfa, who are portrayed as loyalists who flock to ballot boxes in large numbers during elections. This has sparked controversy, prompting mocking discussions and a rise in hate speech on social media.

On the other hand, the page ‘Nael’ reported the ranking of the province of Djelfa based on the percentage of participation in elections in recent years, as published on the Djelfa Info website. With 58 percent, the province came 17th nationally, contrary to the stereotype.

The rhetoric surrounding the inhabitants of Djelfa and Kabyle intensified alongside the 2019 ‘popular Hirak’ movement (popular uprising) and the emergence of pages inciting hatred against them.

Paying for Hatred: Sponsored Campaigns Fuel Regional Rhetoric Online

During the Algerian elections on September 7, 2024, there was an increase in regional rhetoric and polarising language on social media. This discourse was not impulsive or an automatic reaction, but a repetitive pattern whenever elections are held.

Worryingly, this discourse not only circulated through ordinary people’s posts, but was also emphasized and amplified by paid ads on platforms such as Facebook, Instagram, TikTok, and YouTube. Although these platforms claim to protect users from hate speech, they did not effectively impose any restrictions on ad content, allowing polarising, and sometimes racist, messaging to be disseminated in exchange for financial gain.

To confirm the role of these platforms in spreading regional strife and violence, this investigative journalist conducted a digital field experiment in collaboration with a team of digital experts: Ferial M., project manager at Civitech; Asmaa A., a student at the Higher National School of Computer Science (ESI); Sara B., a PhD student at ESI; and Ahmad S., an independent programmer. The experiment aimed to monitor and track ads and posts containing polarising regional language over two months before and after the elections.

During the monitoring process, the team documented dozens of sponsored advertising campaigns that included polarizing terms, some of which incited hatred against specific political or regional groups. Facebook and Instagram permitted the publication of these ads without any objection despite them containing a discourse that violates their community standards.

The terms used were classified according to the groups that used them (regime supporters-regime opponents- sarcastic/neutral discourse), the scale of their dissemination before and after the elections, their use in sponsored ads, and the geographic locations they targeted the most. The study revealed a clear escalation in the use of “regional terms and wording” after the elections, particularly in big cities.

Some terms, like “blue finger” (بوصبع لزرق) and “zouave” (reference to those who enlisted in the army during the French colonial rule) were inserted in sponsored ads, which contributed to amplifying and spreading them. Other slogans, such as “civil and not military state,” also circulated widely through comments and political discussions.

The study concluded that social media platforms did not enforce any meaningful restrictions on the spread of terms targeting specific regions. It also found that paid ads were a key factor in amplifying these terms, suggesting a “lenient” approach by social media platforms in exchange for financial gain, despite the potential social damage.

Djelfa Targeted by Digital Bullying: An Analysis of the Electoral Landscape

To assess the truth of electoral participation in Djelfa province, this investigative reporter travelled to the area on September 7, 2024, to cover the elections, and to find out how people feel about hate speech on social media . The reporter also tried to explore their views on the bullying and polarisation created by the algorithms online.

The reporter visited six polling stations: Megwas Belkacem middle school, Fudeili Abdul Qadir middle school, Belkhiri Mbarka primary school, Si Sharif Ben Lahrache high school, 18 February middle school and Hinichi Mohamed primary school. Although voter turnout was low in the morning, it increased in the afternoon.

He spoke to some young men about electoral participation. Some of them noted the current lack of engagement among young people, whereas voter turnout in Djelfa was high under Bouteflika, fuelled by promises to turn it into a capital city. However, young people today are considering emigrating.

Other young men who participated in the elections confirmed that they were voting to preserve the country’s peace and security, and to enable the president to build on his achievements during his first term in office.

Regarding hate speech on social media, one young man observed that residents were uninterested in these accusations. A polling station worker described the people of Djelfa as peaceful, suggesting that regime opponents were likely responsible for the bullying over the high voter turnout.

One citizen explained that the region usually sees the highest voter turnout during municipal elections because of the influence of the “tribal system”, whereby voters support candidates from their own clan.

The Law on Combating Hate Speech in Algeria

In light of the increase in hate speech in recent years, Algerian legislators enacted the Law on the Protection from and Combating of Discrimination and Hate Speech in 2020. Algeria's Law No. 20/05 defined “hate speech” as “All forms of expression that publish, encourage, or justify discrimination, as well as those that include a style of contempt, insult, hostility, hatred, or violence directed at a person or group of people based on sex, race, color, lineage, national or ethnic origin, language, geographic affiliation, disability, or health status.”

The law stipulates the establishment of a National Observatory for the Prevention and Combating of Discrimination and Hate Speech. The Observatory's aim is to monitor and track such occurrences, and to propose effective prevention measures.

Algeria signed the International Convention on the Elimination of All Forms of Racial Discrimination in 1966 and ratified it in 1972.

On May 1, 2025, the state prosecutor ordered the pre-trial detention of the academic Mohamed El-Amine Belghith, following his controversial remarks on Amazigh identity. The authorities deemed his comments to be an act targeting national unity and inciting hate speech.

Since 2018, incitement campaigns have escalated on social media, targeting residents of Kabyle areas. Several pages have promoted a discourse of discrimination and exclusion against the area’s inhabitants, by exploiting political positions or linguistic and cultural differences.

Naima Salhi, president of the Equity and Proclamation Party, is one of the prominent figures who have stirred controversy in this regard. In her posts and videos, Salhi stated that she would never allow her daughter to speak Tamazight, the language of the Amazigh people. She also called for a boycott of those who flew the Amazigh flag during the Hirak movement.

Before her controversial 2019 comments about the Amazigh flag, Naima Salhi had already faced harassment, threats, and heated confrontations in 2018 over her earlier statements.. She later clarified that she was referring to the “Zouaves”, not Kabyle tribes.

The term “Zouave” refers to Algerian locals who were conscripted into military units formed by colonial France during its occupation of the country, in an attempt to bolster its forces. The Zouaves were a light infantry regiment first established in Algiers, composed not only of tribal members, but also of residents of the city from various origins. According to the historian Mohamed Arezki Ferrad, the term is sometimes used to reveal the true functional role of these troops as “mercenaries” employed by France to further its colonial ambitions.

An Algerian court sentenced Naima Salhi to six months in prison and a fine of 50,000 DA on charges of compromising the nation’s security and for the offense of publishing posts that could undermine the national interest.

Commenting on the motives behind these campaigns, journalist Mohamed Iwanougan explained: “For more than a decade, countless documents, videos, and articles have circulated portraying Kabyle regions and their residents as traitors, holding them responsible for various tragedies that have afflicted the Algerian people.

Djallal Bouabdallah, an expert in digital transformation and cybersecurity, says that the escalation of hate speech on social media, especially during sensitive political periods in Algeria, raises a critical question about the role of algorithms in amplifying and facilitating the spread of this kind of discourse.

Bouabdallah explains that social media algorithms are designed to promote content that receives a lot of engagement, regardless of its nature. Emotional or provocative content usually receives high levels of engagement, which facilitates the widespread dissemination of hate speech, as it provokes anger and division. He said that algorithms do not grasp the concept of “ethics”, but are only concerned with assessing numbers. Therefore, they amplify this discourse unintentionally.

Bouabdallah views the design of the algorithm as a compass that directs content towards specific user groups. When based on a user’s previous preferences and past behaviour, the algorithm tends to limit them within an “information bubble” that reinforces their biases. Consequently, they may regularly be exposed to hate speech or misinformation, particularly if this type of content is popular within their network.

Bouabdallah points out that distinguishing between freedom of expression and hate speech requires an understanding of the cultural and linguistic nuances that algorithms currently lack. Therefore, platforms must rely on a combination of AI and human review, as well as collaborating with local experts and civil society organiszations, in order to understand local sensitivities.

He added: “Algorithms are often better suited to global languages such as English, and struggle to understand local dialects or cultural expressions in languages such as Arabic, Amazigh, and even French with a North African accent. This shortcoming allows the proliferation of hate speech, which is not always detected or punished in the same way as it is in other linguistic contexts.”

On the other hand, Bouabdallah emphasises that tech companies' responses vary depending on the market and its economic or political power. Like many other African and Arab countries, Algeria is not prioritized in these companies' algorithmic moderation strategies. Their efforts are often limited or delayed, which makes the communications environment more vulnerable to infiltration by hate speech or political and societal manipulation.

The cybersecurity and digital transformation expert believes it is necessary to transform moral responsibility into a legal obligation. This would force these platforms to consider local contexts, strengthen monitoring and reporting mechanisms, and ensure that AI is not used to promote social division.

Finally, the investigative reporter's documentation of recent coverage of the Algerian presidential election and the findings of digital monitoring tools confirm that hate speech in Algeria is no longer simply a natural by-product of political and social tensions. Instead, it has become targeted and amplified through digital tools — a result of algorithmic design that favours interaction and profit over social security.

While Algerian authorities have passed legislation to curb hate speech and prosecuted individuals responsible for spreading it, a pressing question remains: what legal measures should be taken against the companies that own these platforms, especially when some of their documented practices fall under Algerian jurisdiction by targeting citizens within the country’s borders?

Overlooking or selectively addressing this kind of discourse only accelerates its normalization. Efforts to fight hate speech often target those who disseminate it, while the true instigators escape accountability.

We contacted the META team for comments but did not receive any response as of the date of publication of this investigation.

This investigation was undertaken with support from ARIJ