Violent Non-State Actors and Generative AI in Warfare: The Rsf and the Sudanese Civil War
Generative AI is rapidly becoming a critical component of information warfare. A number of violent non-state actors are beginning to embrace this new technology, which has substantially altered the information ecosystem within the ongoing Sudanese Civil War.
Introduction
The information sphere has been an important aspect of human warfare for millennia. The ancient Sumerians built massive monuments of their kings destroying their enemies, the pharaohs of Egypt spread state iconography of the “miserable Asiatic” to foment xenophobia among their subjects, and the First Crusade was inflamed by rumors of Muslim atrocities in Jerusalem. With the development of the Internet and artificial intelligence (AI), the information sphere has taken on an even more crucial role in warfare.
Africa is uniquely vulnerable to AI-generated content. Coordinated disinformation campaigns have quadrupled across the continent since 2022, fueled in large part by Russia and China. Russia’s former Wagner Group, now known as Africa Corps, has played a large role in providing AI-generated imagery to the continent, including countries such as Burkina Faso, Mali, and Niger.
Violent Non-State Actors and Artificial Intelligence
As AI continues to reshape the world, a number of terrorist organizations, also known as Violent Non-State Actors (VNSAs), have been quick to adapt and employ this new technology. Although a 2021 conference held by the United Nations concluded that there was no definitive evidence to suggest that terrorist organizations had yet successfully employed AI, the global situation has rapidly changed since then. Organizations such as the Islamic State (IS) have employed open-source AI tools such as ChatGPT to enhance propaganda campaigns. After the March 2024 attack on a concert hall in Russia, IS affiliates widely disseminated AI-created propaganda videos across social media in a bid to gain new recruits; other uses include deepfake audio recordings of IS leaders that are rapidly translated into multiple languages and propagated across the internet. Other examples include the use of AI-generated images of wounded or dying infants and children in the Gaza conflict. However, for the most part, the use of AI beyond propaganda and recruitment appears to be largely “aspirational”.
Unfortunately, this will likely not last long. The potential capabilities that AI could provide, from widespread social engineering to autonomous drones to potential chemical and biological weapons development, could vastly increase the impact of small, decentralized, and resource-poor VNSAs. Jama’at Nusrat al-Islam wal Muslimin (JNIM) may already be employing offline AI machine learning to enhance drone strikes across sub-Saharan Africa; IS supporters have already expressed an interest in pursuing further uses of AI; other groups, such as Al-Qaeda, Hezbollah, and Hamas, are also experimenting with the use of AI. Most notably, a VSNA in Sudan has already begun the widespread use of generative AI in warfare – one of the first documented cases in history.
The Sudan Civil War and the RSF
On 15 April 2023, following a prolonged stalemate after the 2019 ouster of President Omar Al-Bashir, Sudan’s fragile government collapsed into civil war. The two main actors in this conflict are the Sudanese Armed Forces (SAF), led by General Abdel Fattah Al-Burhan, and the Rapid Support Forces (RSF), a paramilitary force led by General Mohamed Hamdan Dagalo, better known as “Hemedti”. This conflict – complicated by regional loyalties, ethnic tensions, and outside interference – has been defined by a massive humanitarian crisis affecting millions, brutal fighting, the reported deployment of chemical weapons by the SAF, the deliberate blockade of humanitarian aid, and ethnic cleansing. In just one example, when the RSF captured the city of El Fasher on 26 October 2025 after a 500-day long siege, the paramilitary slaughtered an estimated 60,000 civilians.
Despite its origins as a local militia employed by the Sudanese state for counterinsurgency, the RSF has been quick to embrace new technologies and battlefield strategies, influenced by contacts with the Wagner Group and a maze of other international commercial and military connections. The RSF has widely deployed Chinese-made drones, including long-distance loitering munitions, and even used precision drones as tools of assassination. In one notable event, the RSF attempted to take al-Burhan’s life in 2024 at a graduation ceremony with a precision drone.
Disinformation in Sudan
Disinformation has a long history in Sudan, and can be divided into several distinct phases. During the al-Bashir regime from 1989-2019, disinformation was widely employed as state propaganda designed to both fuel ethnic violence. The Janjaweed militia, the precursors of the RSF, were emboldened by state-sanctioned propaganda to commit mass atrocities during the infamous Darfur genocide. The regime also focused on targeting dissent; the Cyber Jihadist Unit, founded in 2011, was particularly effective as a form of online surveillance used to target activists and suppress information. After the fall of al-Bashir’s government, both the SAF and the RSF often worked together to spread disinformation. Tactics included setting up fake protests to flush out dissenters, the mass use of bots to ban anti-government rhetoric online, and infiltrating anti-government message boards. These methods were remarkably sophisticated, and both sides used outside companies – including an Egyptian firm that provided a troll farm – to secure a grip on online discourse. With the outbreak of war, these two “seasoned disinformation forces” are suddenly at odds, causing an increasingly violent war of information as both sides seek to “overwhelm the narrative of the other”.
The RSF has proven particularly adept at online messaging, with a significant portion of its budget going toward media and public relations. Examples include the RSF circulating pictures of fighters distributing medical aid after they were accused of slaughtering hundreds of patients at a hospital in el-Fasher, and their widespread use of brutal and violent rhetoric – one viral video shows a female RSF fighter calling for other RSF soldiers to rape women to “purify their lineage”. Social media is essential to this strategy, ranging across apps including WhatsApp, Facebook, TikTok, and Clubhouse. Both sides have also employed social media influencers – often acting under a veneer of neutrality while reinforcing certain points – to aid in disinformation goals.
Notably, the RSF appears to be more focused on an international audience; most of their posts are in English, and Hemedti has placed a strong emphasis on visiting international leaders. The paramilitary also has a strong preference for covering up its many atrocities and casting itself as a pro-democracy force fighting against an oppressive regime.
The RSF is also skilled at coordinated online campaigns – after the RSF’s capture of Merowe Airport, dozens of accounts on X (formerly Twitter) suddenly posted identical videos of the victorious troops, all within minutes of one another. This synchronized campaign falsely gave the impression of broad popular support for the paramilitary. Both sides have used similar tactics of flooding social media with coordinated waves of imagery and messages to create the illusion of popular support.
The RSF’s Use of AI
Information operations have long been identified as a force multiplier, particularly in insurgencies. A number of recent publications have noted the military potential of generative AI. AI-generated images may be employed for deception, to sow confusion among the enemy, to degrade the information sphere and manipulate public perception, and potentially manipulate social media on an unprecedented scale. These techniques are already being employed in Sudan.
While disinformation has a long history in Sudan, the use of generative AI is a new phenomenon. Like many other VSNAs, the RSF has been quick to embrace the use of AI. As early as October 2023, an RSF-associated source released a campaign of AI-generated recordings of the former President al-Bashir. These clips, released as so-called “leaked recordings” on a TikTok channel called ‘The Voice of Sudan’, were generated from a mixture of old audio files and news reports, and featured al-Bashir criticizing the SAF’s leader, al-Burhan. This caused a number of al-Bashir supporters to renege on their support of the SAF. This incident is particularly notable because it is “one of the first documented cases of generative AI being used as a weapon of war.” Other alleged “leaked recordings” of politicians criticizing the SAF and discussing the possibility of joining the RSF have further sown confusion in the public sphere.
Generative AI has also been used to specifically target American interests in the region. In April 2023, another “leaked recording” was released, claiming to be from the U.S. Ambassador John Godfrey. In the recording, an AI-generated clone of Godfrey’s voice outlines a plan to increase secularism within Sudan. While the exact origin of the clip remains unknown, it was likely created by Islamist-linked networks to undermine Western influence within the region.
In the following months, RSF and RSF-affiliated sources continued to employ generative AI. In March 2024, another audio file that claimed to be a recording of al-Burhan ordering the mass killing of civilians was released online. The recording was viewed over 230,000 times, and so popular that it was shared by multiple Sudanese politicians. The next month, an AI-generated image of a building in Al-Jazeera University, allegedly bombed by the SAF, went viral on Facebook. Again, it was shared by many political and public figures across Sudan.
Interestingly, the SAF has attempted to take advantage of the RSF’s use of AI; at one point, the military claimed that all the recordings of Hemedti were AI-generated and that the paramilitary leader was actually dead. While these accusations were patently untrue, they are yet another example of the constant disinformation war waged by both sides.
The SAF also appears to have begun to employ AI images; after the recent fall of el-Fasher, a number of AI-generated videos claiming to be from the city went viral. In one, a mother and her children are cowering in fear, supposedly in front of two RSF troops about to execute her; in another, an elderly woman stares directly into the camera as she sobs. Another video shows a crowd fleeing a burning village, as a woman calls out, “RSF has entered our neighborhood and started killing people.” Other videos include scenes of begging Sudanese children – the most popular of these has received nearly 40 million views on TikTok. While these videos are highly convincing, fact-checkers have determined that they are false. The SAF has even followed the RSF’s lead in AI-generated recordings of political leaders; one pro-SAF group called ‘Sea and River’ has circulated videos of tribal leaders from Darfur claiming to support the RSF, further inflaming ethnic tensions.
AI in an Information Vacuum
At the start of 2024, only 28.7% of the Sudanese population had internet access. With such limited access to the online information sphere, it initially seems counterintuitive that the RSF would invest so heavily in AI-generated content.
However, Sudan is effectively in an information vacuum. Independent Sudanese media has largely collapsed in the past years, with hundreds of journalists forced to flee the country. The online presence of the populace is now increasingly split between supporters of the RSF and the SAF, further fueling ethnic tensions and sectarian violence. Hashtags such as “Truce is treason” have further limited any option for peaceful dissent in the increasingly radicalized environment. The areas that do have internet access are heavily restricted – for example, settlements under the RSF’s control may be able to gain internet access by using communal Starlink devices. However, these devices require an hourly fee and are constantly surveilled by RSF soldiers, allowing for tight information control. Internet shutdowns are common, making the internet closer to a “valve” that the paramilitary is able to shut off at will.
With the lack of any reliable media, tightly regulated internet access, and the chaos of war, rumors now act as the primary source of information in the collapsing public sphere. Both the RSF and the SAF are well aware of this and have specifically designed their disinformation to rapidly spread offline through word-of-mouth. Through this method, digital warfare, even when deployed in a limited space, has directly contributed to physical violence and further radicalized the population.
Implications for American Interests
Unlike Russia and China, the US does not currently have a well-defined strategy regarding information warfare, and has traditionally followed a more defensive posture using such tactics. However, a number of US agencies are currently pursuing the use of AI to identify and counter hostile information campaigns. In 2023, the National Defense Authorization Act set aside $20 billion for AI research and development. The current Secretary of War, Pete Hegseth, has also strongly emphasized the increased use of AI in the Department of War.
This increased research and development of AI may offer a potential counter to the enemy use of generative AI; as we have shown, it is already being deployed to directly counter American interests within Sudan. It is almost certain that this tool will be used against American interests even more in the future.
Conclusion
While information warfare is as old as conflict itself, the use of generative AI is a completely new phenomenon. VSNAs across the world are quickly embracing this new technology, which has already been widely deployed within the information sphere of the ongoing Sudanese Civil War. The potential that this technology has to influence a populace, even in an environment with minimal internet access, should not be underestimated.
Generative AI has already been directly used to counteract American interests in Sudan. This threat will likely only exponentially increase in the future; American policy will need to adapt to counter the use of generative AI by geopolitical rivals.
The post Violent Non-State Actors and Generative AI in Warfare: The Rsf and the Sudanese Civil War appeared first on Small Wars Journal by Arizona State University.