top of page

AI on the Battlefield: Yemen's Civil War

Writer: Grimshaw Club
Grimshaw Club
6 minutes ago
7 min read

This briefing examines the growing role of artificial intelligence in Yemen's civil war, exploring how the Houthis' reported use of AI-assisted weapons development and deepfake technology could reshape asymmetric warfare. It also considers the wider implications for regional security, global shipping routes and the balance of military power, highlighting how increasingly accessible technologies may empower non-state actors while raising urgent questions about AI governance and the future of conflict. This article was written by Blair Shang and edited by Mariana Quevado.


Introduction


On 5 October 2026, Rashad al-Alimi, chairman of the Saudi-backed Presidential Leadership Council (PLC)—the de facto, internationally recognised Yemeni government—formally announced the beginning of Operation Dawn of Yemen, a full-scale military counteroffensive against the Houthi-controlled northwest. The operation was described by al-Alimi as the “decisive phase” in Yemen’s history, aimed at restoring state authority over Houthi-controlled territories after the movement's extraordinary successes on the battlefield, which, as we shall see, were enabled by emerging technologies.


The Yemeni government forces advanced along the coast in southwestern Yemen near cities such as Taiz, Marib, and the capital Sanaa. More than 100 Saudi aircraft supported the Operation, striking hundreds of Houthi targets. In retaliation, the Houthis launched massive missile and drone attacks across a number of Saudi cities, targeting transportation hubs and energy infrastructure, such as the King Khalid International Airport, including a deadly attack in October that killed at least 18 people and injured more than 300. 


Since the start of Yemen’s Civil War in 2014, the Houthis, or formally known as Ansar Allah, have controlled Yemen’s capital Sanaa and much of the country’s population. The rebels have received substantial Iranian support, including weapons, funding, and military training, violating—in spite of Tehran’s repeated denials—a 2015 UN arms embargo imposed on the Houthis. With support from the Iranian Islamic Revolutionary Guard Corps (IRGC) and Lebanese Hezbollah, the movement developed increasingly sophisticated weapons, such as ballistic missiles, drones, and explosive-laden fast ‌boats, bolstering its capacity to threaten maritime traffic and disrupt shipping routes. 


In October 2023, the Houthis launched missiles and drones against international shipping in the Red Sea, particularly towards merchant vessels transiting the Bab al-Mandab Strait, triggering a maritime crisis. In response, an American-led multinational coalition, with the participation of more than 20 countries, launched Operation Prosperity Guardian in late 2023 to protect maritime traffic. However, in spite of extensive efforts, including 931 airstrikes against missile and drone launch sites across Houthi-controlled Yemen, at the cost of nearly $7 billion over the course of hostilities, the coalition had simply failed to deter and eliminate Houthi capabilities. Meanwhile, shipping costs through the Red Sea surged by 250 percent, forcing major companies to reroute, and threatening global supply chains. 


On 3 September 2026, after a sequence of intensifying skirmishes, the Houthis launched a surprise offensive against the PLC. Fighting began in the Taiz governorate, which had been besieged by the Houthis since 2016, prompting the PLC to initiate counter-offensives towards Al-Jawf and Al-Bayda. However, following a decisive battle over Mokha—a strategic port city along the Red Sea—the entire west coast of Yemen, with the sole exception of the Midi district, fell under Houthi control. In turn, the Houthis have named this campaign: Operation Allah is Stronger in Might and More Severe in Punishment. 


Yet, several questions remain over the Houthis’ recent successes. How did the group develop advanced anti-ship and ballistic missile capabilities in spite of decades of sanctions? And moreover, how did they achieve such rapid breakthroughs within the span of days, after more than a decade of protracted military stalemate? While Iran’s extraordinary support could help explain the former, and the element of surprise and military preparation the latter, another factor may have played a crucial role within these events: artificial intelligence (AI), signifying an emerging dimension to contemporary warfare.


International Implications


In September 2026, Anthropic revealed that a “weapons engineering cell” operating from the Houthi-controlled northern Yemen had used Claude AI to assist in developing three advanced weapons programmes. The report itself provided a rare, documented case of current large language models being incorporated into active weapons-development efforts, not least by non-state groups.


The use of AI reflects a shift towards domestic weapons development and indigenous design, reducing the Houthis’ reliance on Iranian support. Andreas Krieg, a senior lecturer at King's ​College London, explained that “former Yemeni military engineers and a growing generation of locally trained specialists have helped institutionalise these capabilities… [in essence] Tehran has not simply supplied the Houthis with weapons. It has helped create an indigenous military-industrial learning process that ​becomes progressively harder to dismantle, even if direct Iranian supply routes are disrupted.” The Houthis have therefore emerged as Iran’s most strategically valuable and cost-effective regional military proxies, since their evolving technical expertise and domestic production capabilities become harder to eliminate through military strikes alone. 


Moreover, as Houthi forces advanced towards Mokha, a recording had allegedly spread among Yemeni troops depicting Lieutenant General Tareq Saleh, nephew of former President Ali Abdullah Saleh and the commander of UAE-backed National Resistance Forces (NRF), announcing a withdrawal. Military media later publicly disputed the video, claiming it was fabricated using artificial intelligence and falsely attributed to Saleh to confuse soldiers, influence perceptions, and damage morale. But the damage had already been done. Mokha had served as the main base of Lt. Gen. Saleh, and the loss of the city would become both a military and symbolic blow to the anti-Houthi camp.


In the aftermath of the Houthi offensive, the Yemeni government lost substantial territories around the Bab al-Mandab Strait, including the nearby islands, giving the Houthis newfound leverage over the strategic international shipping route. Simultaneously, that advance opens an additional front in the wider US-Israeli war launched in February against Iran, to the other side of the Arabian Peninsula. Moreover, the offensive exposed the fragmentation of the anti-Houthi camp, which included a variety of autonomous armed groups that were previously supported by different regional patrons and had competing political ambitions. Despite Saudi attempts to bring them under a more coherent, unified command structure, it has not erased years of fragmentation.


Since Iran’s closure of the Strait of Hormuz, Saudi Arabia has sought to redirect its oil exports via alternative maritime routes, like the Bab el-Mandeb Strait. The Houthis’ recent successes dashed such hopes, instead posing a twofold threat to Saudi Arabia. Historically, Riyadh has long faced the threat of Houthi attacks on its energy infrastructure, partly in response to its continued support for the PLC; and, expectedly, the Houthis have used the present opportunity to declare an oil embargo targeting Saudi Arabia.


The mounting pressure to safeguard critical energy export routes has reinforced the need for closer cooperation between Saudi Arabia and the United Arab Emirates (UAE), despite preceding animosity: previously, in January 2026, the Saudi air force bombed a UAE arms shipment bound for Yemen's Southern Transitional Council (STC), a separatist group in opposition to the PLC. Recent developments, however, suggested a rapprochement between Saudi Arabia and the UAE, where the STC’s Southern Giants Brigades coordinated with the Saudi air force in the wider offensive against Houthi-controlled positions.


In the near future, any resolution to the Red Sea Crisis could be a bilateral agreement, a regional security arrangement, or a broader international settlement. However, several preconditions remain: will the Gulf states reach a consensus and coordinate their responses? Could Saudi Arabia mobilise the Mecca Defence Alliance following the Houthis’ recent attack on Mecca itself? And would the United States be able to withstand the mounting pressure from rising fuel prices until the end of midterm elections? Ultimately, the prospects for a settlement will depend on how regional and international powers weigh their competing interests and whether they can coordinate their responses.


On the other hand, the usage of AI in warfare has yet to provoke the public outcry necessary to drive the establishment of meaningful industrial standards and institutional safeguards. The military application of AI remains fraught with legal and ethical uncertainties, whereas evidence of their broader consequences is still developing. What is clear, however, is that governance has remained insufficient at both the national and international levels to regulate these developments and constrain their potentially destabilising effects.


Conclusions


The Houthis’ recent military gains raise broader concerns over the vulnerability of critical maritime chokepoints and the future of regional energy security. Their ability to threaten shipping through the Bab al-Mandab Strait could expose Saudi Arabia and other Gulf states to renewed economic and strategic pressure, particularly if disruptions to energy exports coincide with wider regional instability. Meanwhile, the US reluctance to deepen its military involvement amidst domestic political pressures, such as the midterms, creates opportunities for Iranian proxy groups to escalate or expand their operational reach. Subsequently, the Houthis have exploited this strategic opening by targeting Saudi Arabian Oil Company (ARAMCO) production facilities as well as the PLC’s presidential palace complex in Aden, marking a significant escalation of the present conflict.


More broadly, the reported use of AI-generated deepfakes to mislead PLC forces also points to an emerging form of psychological warfare, echoing Russia’s usage of disinformation operations during the annexation of Crimea in 2014 and the opening hours of the invasion of Ukraine. The significance of AI-enabled disinformation extends beyond false information. Traditionally, electronic warfare has focused on disrupting or denying an adversary’s communications and access to information. Instead, AI-generated deception introduces an additional dimension by actively flooding the combat environment with convincing, fabricated material, undermining information validity and discernment.


As AI systems become more capable, the potential for such operations to disrupt command structures and exploit uncertainty may grow. They could wholly undermine confidence in intelligence, disrupt decision-making, and make it harder for military and political leaders to distinguish reliable information from deliberate deception, signalling a broader shift in the relationship between AI, military power, and technological self-sufficiency. 


Most importantly, AI-enabled tools allow small states and non-state actors alike to develop relatively inexpensive weapons capabilities with effects disproportionate to their cost. The proliferation of low-cost drones and ballistic missiles, for example, coupled with AI, lowers barriers to developing increasingly sophisticated military capabilities, such as the reported use of commercial AI models by the Houthis. This process could possibly and fundamentally change the balance between major and small powers. As the cost of resisting falls and the potential cost of occupation rises, outright military victory and the subjugation of smaller states may become increasingly difficult to achieve without incurring unsustainable losses, as presented by the American hesitation to directly deploy troops in Iran and Yemen. 


In the long run, this change could create stronger incentives for diplomacy, since even militarily superior states face greater constraints on their ability to impose their will by force. Nonetheless, this outcome is not guaranteed. The same technologies that foster peace through deterrence can also fuel future arms races, encourage pre-emptive strikes, and make conflicts more difficult to contain. Moreover, the closer integration between AI development and defence institutions raises questions about the future direction of AI research. If commercial investment and funding pressures encourage AI companies to pursue defence contracts, then military applications may become an increasingly important driver of innovation.


The resulting danger is not simply that the major states will acquire more advanced weapons, but that increasingly destructive and sophisticated capabilities will become accessible to rogue, non-state actors. Ultimately, whether technological developments will produce a more peaceful international order depends on whether the growing costs of military domination can translate into meaningful incentives for restraint and diplomatic compromise.

 
 
 

Comments


LSE's flagship student organisation for International Relations since 1923

Get in touch: grimshaw@lsesu.org

bottom of page