Operational Errors and Civilian Casualties
Abstract
This article examines how inadequate intelligence gathering, operational errors, and a lack of transparency in U.S. military strategy during the Afghanistan war contributed to civilian casualties. Focusing on the transition to aerial warfare after 2014 and covert night raids, it highlights how reliance on pattern-based targeting, limited accountability, and covert operations undermined civilian protection, fueled insurgent propaganda, and eroded legitimacy. The piece advocates for stronger safeguards, independent review mechanisms, and clearer accountability frameworks to mitigate civilian harm in future conflicts, particularly as artificial intelligence and autonomous systems increasingly shape modern warfare.
Two Decades of War
In the wake of the September 11 attacks, the United States launched the Global War on Terror to dismantle al-Qaeda. Central to this campaign was toppling the Taliban regime (1996-2001), which had provided sanctuary to al-Qaeda, leading to a 20-year-long military engagement in Afghanistan. By the time the U.S. forces withdrew in 2021, the conflict had cost approximately $2.3 trillion. The human cost has been as staggering, with an estimated 174,000 people killed. Despite the massive investments of resources and human life, the Taliban returned to power in 2021, and al-Qaeda remains active in Afghanistan, rebuilding quietly while maintaining close ties with the Taliban.
In response to the conflict’s outcome, numerous studies have been conducted on the war, citing factors such as inconsistent decision-making, corruption, strategic miscalculations, and the inability to disrupt insurgents’ financial networks, particularly through flawed counternarcotic programs. To consolidate insights from the war, the Special Inspector General for Afghanistan Reconstruction (SIGAR) launched the Lessons Learned Program. Congress later expanded this effort by establishing the Afghanistan War Commission to examine the full scope of U.S. military, diplomatic, and intelligence involvement.
Building on these assessments, this article examines two key missteps in the war’s execution: first, lethal operational errors resulting from inadequate intelligence gathering and insufficient analysis; and second, a lack of transparency. These missteps are analyzed in the context of the 2014 transition from ground operations to aerial warfare and the conduct of covert night raids. These failures had severe consequences for civilians, shaping public perception and the war’s trajectory. While some argue that fewer operational constraints might have brought the conflict to a quicker end, they often result in civilian harm, alienating local populations and enabling insurgent propaganda.
The Promise and Peril of Precision Warfare
In 2014, the United States declared the end of its ground combat operations and shifted to primarily providing air support and advisory assistance to Afghanistan forces battling the Taliban, while continuing covert operations. Two years later, in 2016, the use of airstrikes was authorized against the Taliban, expanding beyond prior limitations that allowed strikes only in cases of self-defense when U.S. troops came under attack. This transition marked a broader transformation in the U.S. military’s approach to warfare; one increasingly characterized by the use of remote-controlled aircraft and drones operated by people thousands of miles from the battlefield, relying on suspicion-driven targeting. This model of high-tech warfare has since become a defining feature of military engagement in war zones, and it is poised to shape future conflicts.
However, this strategy proves deeply problematic in an asymmetric conflict where non-state armed actors easily blend in and operate within civilian areas, blurring the boundaries between military and civilians. In such environments, infrastructures often turn into dual-use objects. Under such conditions, accurately identifying and striking targets becomes increasingly difficult, particularly in urban centers.
The Taliban in Afghanistan exploited this operational ambiguity, often using civilians and their homes as shields. Their frequent use of Lower-Case Improvised Explosive Devices (IEDs) – including roadside bombs and suicide attacks – indiscriminately killed civilians and military personnel alike. In some cases, they used children as young as 13 to carry out these attacks – further complicating the distinction. ISIS-K emerged in 2015 in eastern Afghanistan, adding to the already existing toll of civilian casualties through the same tactics. Between 2011 and 2016, IEDs accounted for approximately 81% of all civilian casualties. International humanitarian law requires strict adherence to the principles of distinction and proportionality. Nonetheless, the Taliban and ISIS-K routinely violated these principles.
The United States increased its reliance on aerial operations after transitioning away from ground combat in 2014. Airstrikes became the centerpiece of the U.S. strategy. In this transition, minimum civilian harm was a hoped-for outcome. However, in practice, according to an investigation, in Afghanistan, Iraq, and Syria, strikes were often dependent on the quality of surveillance, behavioral cues, and pattern-based targeting, such as vehicle movement or location patterns, rather than reliable, human-sourced intelligence. This approach increased the risk of error, especially in urban environments where civilian presence was unavoidable. As a result, 70–90% of those killed or injured by air-dropped munitions in these settings are civilians.
The scale of air operations was extensive. The U.S. Air Force Central Command conducted 26,227 airstrikes between 2014 to 2021 in Afghanistan. However, these figures have been contested, which has since raised concerns about transparency in the conduct of war. For instance, in 2016, at least 456 airstrikes were not recorded as part of the open-source database maintained by the U.S. The North Atlantic Treaty Organization (NATO) released monthly airstrike data until 2015, but reports grew increasingly inconsistent, with significant discrepancies between the U.S. government, United Nations Assistance Mission in Afghanistan (UNAMA), and independent media. In addition to the number of airstrikes, the ones that resulted in civilian casualties have not been adequately assessed.
Several high-profile incidents illustrate the human cost of these operations. In 2015, a U.S. airstrike struck a Doctors Without Borders trauma center, killing at least 42 medical staff, patients, and caretakers. The center was not considered a dual-use facility and had no military function. A subsequent U.S. investigation acknowledged that 30 civilians had been killed and 37 injured, citing “avoidable human error” as the cause. Contributing factors included operational fatigue, inadequate planning, equipment failure, and the absence of effective safeguards.
In 2017, another controversial joint operation led by Afghan and US-led NATO forces in Kunduz province resulted in airstrikes that reportedly killed scores of civilians. While UNAMA reported 10 civilian casualties, local sources estimated as many as 55. Conflicting accounts have emerged about how civilians came under fire, ranging from claims that they were forced by the Taliban to retrieve bodies to reports that they were returning to assess damage or evacuate casualties. The U.S. military declared the result of its investigation and concluded that “no evidence of civilian casualties has been found.”
In targeting ISIS-K strongholds in eastern Nangarhar province, U.S. forces conducted heavy bombardment, including the use of GBU-43/B, the U.S.’s most powerful non-nuclear bomb, in 2017. For numerous other air operations in the area, Human Rights Watch interviewed local residents who reported that at least nine civilians fleeing ISIS-K-controlled territory were killed in their private vehicle, and that an airstrike on a funeral killed nine and injured eight people, including women. No formal post-strike investigation was conducted. Due to security reasons, the U.S. forces mostly did not conduct on-site investigations after the strikes and mostly relied on satellite imagery and reports provided by the Afghanistan security forces.
After a deadly ISIS-K suicide bombing at Kabul’s airport during the 2021 U.S. withdrawal, which killed more than 100 civilians and 13 U.S. service members near the Abbey Gate entrance, a retaliatory drone strike intended to prevent further attacks instead killed ten civilians, including seven children and a humanitarian worker. The strike was based on an eight-hour pattern-of-life analysis, surveillance, and threat assessments, but ultimately misinterpreted the target’s behavior, which underscored the limitations of the strike in high-stakes environments. The U.S. officials claimed that poor visibility and misinterpretation of the target’s driving behavior contributed to the mistaken identity.
The cases are not unique to Afghanistan. Similar patterns have occurred in Iraq and Syria. In Ramadi, Iraq, a child was unintentionally killed after being perceived as “an unknown heavy object.” And in Raqqa, Syria, at least seven civilians were killed after the Coalition forces incorrectly assessed that the individuals were carrying weapons and wearing tactical vests, even after six hours of observation.
These patterns of error reveal a critical gap between the technological promise of precision warfare and its operational reality. While proponents argue that modern surveillance and targeting systems uphold humanitarian standards, evidence suggests that civilian safety remains highly vulnerable. Furthermore, the Department of Defense’s insufficient efforts to strengthen its capabilities to assess civilian risk through improved intelligence, battlespace awareness, and analysis of strike outcomes, undermine opportunities to learn from past mistakes.
There is growing enthusiasm about the use of artificial intelligence in military targeting, as it can quickly analyze large volumes of data from satellites, signals, human intelligence, and open sources to support precise decision-making. However, the effectiveness of AI is contingent on the quality of data it processes. If the data are biased or flawed, such as the example provided above, it can result in targeting errors and increased civilian casualties, particularly in dual-use facilities and densely populated urban areas.
CIA Raids and the Erosion of Accountability
Beyond airstrikes, CIA-led paramilitary operations further blurred the lines between combatants and civilians. In the war’s early days, the U.S. detained many individuals suspected of al-Qaeda links, with some subjected to torture or held without due process. As the Taliban regrouped in Pakistan, they began recruiting fighters from border areas and sending them back into Afghanistan. These insurgents blended into civilian communities, planting roadside bombs and conducting suicide attacks.
The U.S. adopted more covert methods of engagement, turning to nighttime “kill or capture” raids, operations trained and supported by the Central Intelligence Agency (CIA), which expanded significantly following the major withdrawal of U.S. troops in 2014. A set of elite Units, known as 01, 02, 03, and 04, were created under the Special Forces of the National Directorate of Security (NDS), Afghanistan’s intelligence agency.
These raids were often conducted at night in residential neighborhoods suspected of harboring the Taliban. Because of the covert nature of these missions, limited formal information exists regarding the Units’ structure and operations. Most of what is known comes from investigative journalists and human rights organizations. Over time, these operations became notorious for causing civilian harm and unlawful detentions, leading to widespread public backlash. Individuals were frequently targeted based on tenuous or presumed ties to insurgents or due to their proximity to known suspects. The Taliban capitalized on these incidents, producing and circulating videos that highlighted civilian casualties from night raids to bolster their narrative and gain local support.
Investigations by the United Nations and human rights organizations documented serious abuses, including extrajudicial killings, forced disappearances, and potential summary executions. A 2019 UNAMA report recorded multiple incidents of civilian harm during such operations, while Human Rights Watch identified 14 cases of grave human rights violations by the Units between 2017 and 2019.
Victims and their families were often left without any recourse. There was no official contact point to seek accountability when missions went wrong. Because the Units operated under CIA oversight and not as conventional military forces, they fell outside the legal frameworks that govern international military operations. Subjecting them to Afghan jurisdiction would have conflicted with the CIA’s covert mandate, leaving a legal and accountability vacuum.
These practices not only potentially breached international humanitarian law but also undermined the legitimacy of the Afghanistan government and alienated local communities in areas targeted by the raids. Throughout the 20-year conflict, CIA-sponsored operations, particularly the night raids carried out by Afghan paramilitary Units, remained shielded from public oversight.
Toward a More Accountable and Humane Warfare Framework
Moving forward, it is essential for the United States to enhance the accuracy of its intelligence-gathering with a stronger emphasis on civilian protection. To foster accountability and learn from operational mistakes, an independent post-strike review board should be established to investigate incidents involving civilian harm, evaluate targeting decisions, and recommend improvements. Ultimately, military engagements must be guided by transparency, oversight, and a clear commitment to minimizing civilian casualties, especially in an era where advanced technologies promise precision but too often yield unintended destruction.
Department of Defense’s 2022 Civilian Harm Mitigation and Response Action Plan (CHMRAP), prompted by the 2021 drone strike in Kabul, was a step in the right direction. The plan reads: “Protecting civilians from harm is not only a moral imperative, it is also critical to achieving long-term success on the battlefield.” However, CHMRAP was not applied retroactively, leaving the victims of airstrikes before 2022 without the opportunity to have their cases re-examined for credibility. Also, its future remains uncertain, and concerns persist over efforts to scale it back. In addition, the United Nations’ Political Declaration on Strengthening the Protection of Civilians from the Humanitarian Consequences Arising from the Use of Explosive Weapons in Populated Areas offers additional pathways for reform.
The Afghanistan War Commission and SIGAR’s Lessons Learned Program should provide specific guidelines and statutory measures that outline how their findings will be used to ensure accountability and justice for victims as well, and how they will help the U.S. government avoid repeating similar errors in other contexts.
Finally, civilian protection remains a concern in the age of artificial intelligence and autonomous drones, particularly as the laws of war were originally designed for conventional conflicts. A critical question arises: how can an autonomous drone reliably distinguish between a civilian and a combatant without sufficient or unbiased data? Until now, the U.S. Department of Defense Directive 3000.09, which governs the development and use of autonomous weapons systems, mandates human in-the-loop to ensure accountability, ethical judgment, and compliance with international law. However, the effectiveness of such policies ultimately depends on how they are implemented, especially in complex urban settings where limited transparency and weak data inputs continue to contribute to civilian harm.
