2025 Civilian Casualties Report Reveals U.S. Civilian Protection Gaps
Despite admitting to its biggest civilian harm incidents in years, the U.S. military has not accounted for dozens of victims from 2025.
In late August, the Department of Defense released its annual report on civilian harm from U.S. military operations in 2025, acknowledging that the U.S. was responsible for killing 153 civilians and injuring another 243, in addition to nine deaths and 14 injuries newly acknowledged from previous years. The report demonstrates concerning gaps in the U.S. military’s ability to prevent civilian harm—and a lack of accountability when it does occur.
Often referred to as the “1057” report after the section of the 2018 National Defense Authorization Act (NDAA) that requires the Pentagon to submit this analysis to Congress, the annual report plays a vital role in providing public transparency and enabling congressional oversight related to the human costs of U.S. military operations. Such oversight plays a vital role in building better policies and practices to protect those caught in warfare.
It was congressional pressure and decades of lessons learned from operations in Afghanistan, Iraq, and Syria which enabled the instigation of the Civilian Harm Mitigation and Response Action Plan (CHMR-AP) in 2022, which laid out steps for the military to overhaul its practices for preventing and responding to civilian harm over four years. In 2023, the plan was followed by a policy governing such practices, the Department of Defense Instruction on civilian harm mitigation and response (DODI). Together, this patchwork of Department of Defense policies and congressional mandates formed a critical framework for learning from harm and improving civilian protection efforts.
However, at the onset of the second Trump administration and under the watch of rules-of-war-averse Defense Secretary Pete Hegseth, targeted budget cuts and a stated prioritization of “lethality” have weakened or stalled many of those institutions and practices. In the Department of Defense’s new approach of self-imposed resource constraints for civilian harm mitigation and response (CHMR), the 2025 civilian casualties report provides one of the first insights into how the U.S. military has assessed and addressed harm from its operations under the Trump administration. This is significant at a time when the U.S. has still not directly accepted responsibility for major incidents more recently, such as the strike on the first day of the Iran war that killed more than 150 children.
The report reveals a number of apparent shortcomings and raises significant concerns that merit further congressional and public scrutiny. For instance, the report suggests that the U.S. military now has seemingly little capacity to investigate even a relatively small caseload of harm claims. The report also acknowledges two of the largest mass casualty incidents in U.S. military history, both of which occurred in Yemen in 2025, yet demonstrates little appetite to respond to or learn lessons from such devastating harm.
Unprecedented Harm Acknowledged in Yemen
While the U.S. has been carrying out airstrikes and drone strikes in Yemen since 2002, operations ramped up when Operation Rough Rider was launched against the Iranian-backed Houthi political and militant group on March 15, 2025. Just over a month later, during the evening of April 17, 2025, the U.S. conducted an airstrike targeting the Ras Isa Port in Hodeida, Yemen. In combination with the nearby ports of Hodeidah and Al-Salif, Ras Isa is the entry point for nearly 80 percent of humanitarian aid for Yemen. In the 2025 civilian casualties report, the U.S. acknowledged for the first time that 80 civilians were killed and 171 injured in this incident. At the time of the strike, U.S. officials dismissed allegations of harm as “Houthi casualty numbers,” conflicting with the 1057 report, which acknowledges that “a significant amount of reliable information was available to inform a civilian harm assessment regarding the strike,” including open-source information. The strike represents one of the deadliest incidents ever conceded by the U.S., and it is the highest death toll reported by the Department of Defense since the 1057 report’s creation.
Eleven days after the attack on the port, the U.S. conducted another deadly airstrike on a site where Houthis detained people who had traveled into Yemen as part of a wider migration route toward Saudi Arabia. This was a known migrant detention site, visited by humanitarian agencies on multiple occasions. U.S. officials declined to comment on the allegations of civilian harm until the 2025 civilian casualties report was released over a year after the attack, and it acknowledged the deaths of 68 civilians and the injuries of another 47, numbers that align with those identified by civilian harm watchdog Airwars, where two of this article’s authors work.
These are extraordinarily high casualty numbers. The U.S. has recognized only one comparable incident in the past decade, when a U.S. strike collapsed a building sheltering people during the war against the Islamic State in Iraq in 2017, killing at least 105 civilians. Amnesty International and Human Rights Watch have called for investigations into both strikes as possible war crimes. Given the devastating human toll of these strikes, it is notable that the report did not include details about what went wrong or lessons learned from these tragedies.
Alongside these admissions, the U.S. also conceded a third incident from the early days of the war. On April 6, 2025, just weeks after Operation Rough Rider had been announced, a U.S. airstrike hit a home in the Shaab Al-Hafa neighborhood of the capital, Sanaa. The U.S. acknowledged in the 1057 report that the strike had killed five civilians and injured 25 others.
In the report, U.S. Central Command (CENTCOM) acknowledged that it was “currently” assessing 15 other incidents but did not specify what these incidents were. CENTCOM has a submission form on its website to report civilian harm from U.S. military actions in its area of responsibility—as do U.S. Southern Command (SOUTHCOM) and U.S. Africa Command (AFRICOM)—and it also conducts its own open-source research into civilian harm allegations, meaning CENTCOM receives and assesses many civilian harm incidents. Yet there is little transparency from CENTCOM regarding which incidents are currently being reviewed since they stopped publishing monthly civilian harm releases, which gradually slowed down starting in 2021.
Excluding contested strikes, Airwars tracked 30 additional incidents of civilian harm in Yemen in 2025 that were not mentioned in the 1057 report. This leaves at least 15 incidents of harm seemingly unexamined by CENTCOM. Given the reduction in publicly released information from CENTCOM, it is unclear whether this means these incidents were assessed and dismissed, or were excluded from assessment for another reason.
Finally, when civilian harm occurs, Congress has authorized the Department of Defense to make condolence payments, known as ex gratia payments, to civilian victims and survivors as a way of recognizing and addressing the harm. Yet in each of the acknowledged Yemen incidents from 2025, the Pentagon also stated that “ex gratia was determined to not be appropriate for this incident.” This decision appears premature, as the confirmation of civilian harm typically initiates discussions with affected families about possible amends. Rather than dismissing amends out of hand, the process ideally would have included a good-faith effort to understand families’ needs and preferences as well as the logistical and security challenges to potential responses.
Reviewing Past Harm in Iraq and Syria
Over the past decade, the U.S. military has led a coalition—termed Combined Joint Task Force Operation Inherent Resolve—against the Islamic State in Syria and Iraq. While some of the coalition partners have ceased operations, the U.S. military continues to engage in airstrike and “air drop” operations in Syria and Iraq, which are sometimes announced in press releases issued by CENTCOM. The U.S. did not concede it was responsible for any civilian harm in Iraq or Syria in 2025, but Airwars has identified at least 11 incidents attributed to U.S. forces or the U.S.-led coalition against the Islamic State in Syria throughout the year. Twelve civilians were killed, and at least 11-14 others were injured during the 11 incidents. All 11 Airwars-identified incidents included the names of individuals killed or injured and/or an exact location, information that is highly valued during the investigation process based on declassified documentation of internal civilian harm investigations published by the New York Times for U.S. military operations in Syria and Iraq.
The U.S.-led coalition, also referred to as Operation Inherent Resolve, announced in its July 2025 Quarterly Civilian Harm Report that at least three civilian harm assessments remained open as of June 30, 2025, including two incidents from 2025 and one from 2018. The 1057 report acknowledged the deaths of eight civilians and the injury of 14 others in a 2018 strike in Iraq that mistakenly hit a gathering of Iraqi security forces. Airwars documented deaths and injuries from two families in the attack. While the report makes no mention of a determination about whether to provide ex gratia payments to the survivors, the department should consider ex gratia and other forms of amends in this and other acknowledged cases, in consultation with the affected families and their representatives.
The other two incidents mentioned in the quarterly report from 2025 are not acknowledged or referenced in the 1057 report, and the report is unclear about whether this is because they have been dismissed or are still being assessed.
A Blind Spot in Somalia and Nigeria
The only case of civilian harm acknowledged in Somalia was a single civilian death in El Buur in 2018 that had been reassessed after new information came to light about the incident. This is despite the fact that during President Trump’s first year of his second term, a record-breaking number of strikes were conducted in Somalia against al-Shabaab and ISIS-Somalia. Among the 126 strikes announced by AFRICOM in Somalia in 2025, more than in any other single year since U.S. operations began in 2006, the Department of Defense did not report any civilian casualties in Somalia for 2025. In contrast, Airwars documented three incidents of harm in Somalia in 2025, including a strike near Jamaame in November that killed up to 25 civilians and another in which a respected tribal leader was killed.
The 1057 report states AFRICOM “acted on” six reports of civilian harm in 2025, including two investigations that were still ongoing from 2024. Notably, two reports of harm led the command to open more extensive investigations. The 2025 report does not clearly articulate the status or result of those investigations and instead notes that two of those six cases “acted upon” were dismissed, three remained ongoing, and one was confirmed as harm resulting from U.S. operations (the El Buur incident described above).
Beyond Somalia, U.S. operations in Nigeria in 2025 are absent from AFRICOM’s reporting in the 1057. This is despite reported civilian casualties from a U.S. airstrike in the country on Christmas Day, resulting in the injury of five civilians, including two women and a child, when missiles hit a hotel and other buildings near the intended target. It is unclear if this incident is included in AFRICOM’s list of ongoing assessments or investigations.
Operation Midnight Hammer in Iran
In 2025, the U.S. attacked three nuclear facilities in Iran as part of Operation Midnight Hammer, the first direct U.S. military attack on Iran in decades. The 1057 report briefly mentions these strikes in Iran but claims that CENTCOM assessments found no civilian casualties from the operation. This finding contradicts reports from the head of the Red Crescent, which is part of the International Red Cross and Red Crescent Movement, of injuries from the U.S. strikes on nuclear facilities in Fordow, Natanz, and Isfahan, including 11 injured individuals, four of whom were hospitalized. The specific nuclear site where the injuries took place was not specified.
Latin America and the Caribbean
The U.S. conducted a reported 35 strikes (as defined by Airwars’s methodology) on small boats in the Caribbean Sea and the eastern Pacific Ocean in 2025 as part of Operation Southern Spear, a campaign lethally targeting alleged drug traffickers. These strikes, taken against criminal suspects with no due process outside of any recognized armed conflict, constitute extrajudicial killings of civilians.
In 2025 alone, Airwars documented 123 deaths, including 10 survivors presumed dead and two injuries, from these 35 strikes. The 1057 report, however, claims no civilian casualties in the Western Hemisphere. The report’s conclusion rests on a highly contested legal opinion. The administration has not made its legal rationale for the attacks public, but it appears to assume (a) that the U.S. is in an armed conflict against drug trafficking organizations and (b) that drug trafficking suspects, or the vessels they are on, are lawful military targets.
Yet legal experts across academia, civil society, and the United Nations agree that U.S. operations in the region do not constitute an international or noninternational armed conflict. International human rights law, rather than international humanitarian law, governs the use of force in these strikes, and the U.S.’s civilian versus combatant distinction is entirely inapplicable. The strikes therefore amount to extrajudicial killings and are not lawful military operations. As Charlie Trumbull has argued, they may also constitute crimes against humanity.
Even amid the contested legal interpretations, the 123 civilians killed in boat strikes—by the U.S. military’s own count—are a glaring omission in the 2025 civilian casualties report. This is consistent with the administration’s opacity around the strikes: The Trump administration has repeatedly failed to respond to congressional requests for information about the strikes and has even admitted it does not know the identities of those killed and has not required the military to identify whether the individuals are drug traffickers before the strikes.
Some of the most comprehensive efforts to identify civilian victims are being led by the journalist collective Centro Latinoamericano de Investigación Periodística (el CLIP), whose findings have revealed that those identified have little to no affiliation with criminal organizations. The American Civil Liberties Union is currently challenging the lack of compensation and redress avenues for two families who have lost a loved one in these airstrikes.
Structural Challenges Revealed
Beyond geography-specific gaps, the 2025 civilian casualties report underscores several structural challenges. First, the low ratio of completed-to-outstanding assessments speaks to severe capacity constraints in combatant command assessment cells. As noted above, CENTCOM currently has 15 Yemen reports under assessment. That CENTCOM worked through only three civilian harm assessments between the operation’s conclusion in April 2025 and the report’s compilation in early 2026 suggests the command lacks the capacity to conduct assessments at this scale and tempo. Indeed, the Trump administration has clearly marked this as a low-priority issue, drastically reducing staffing and funding for civilian harm mitigation and response across the U.S. military. The cuts leave CENTCOM with just one individual responsible for processing civilian harm assessments. SOUTHCOM and AFRICOM have similarly low staffing levels, with just one CHMR-dedicated individual at each. AFRICOM’s lack of capacity likely contributed to delays in assessment completions as the U.S. continues its air campaign at unprecedented levels in Somalia.
Additionally, as with many U.S. military operations, higher-profile incidents appear more likely to prompt a response than incidents that garner less media attention in the United States. Yet in a context such as Yemen, restrictions from the Houthis, closed civil society networks, and access limitations mean that international news outlets covered only a handful of incidents.
Finally, while the 1057 report acknowledges the individuals killed and harmed in these deadly airstrikes, the report offers no explanations as to how such significant levels of harm could occur. Nor does the report explain how the Department of Defense will capture and act on the lessons necessary to avoid repeating the same errors.
With almost 300 incidents of civilian harm tracked through open-source research in just the first 40 days of the war in Iran, it is clear that lessons from civilian harm in 2025 were not integrated effectively into Defense Department operations or targeting practice by the time the U.S. launched its war on Iran 10 months later; on the first day of that war, the U.S. conducted another mass casualty event, killing at least 150 civilians in a school, most of whom were children.
While U.S. officials remain silent on civilian harm allegations in Iran, anonymous statements by Pentagon personnel point to several possible causes: Notable among these possible flaws is the role artificial intelligence (AI) played in military targeting, facilitating a newfound speed and scale of warfare. Indeed, this is a lesson that could have been learned from U.S. operations in Yemen, where the use of AI may have facilitated an unprecedented uptick in airstrikes, with 339 strikes conducted between March 15 and May 6, 2025; in the same two-month period, as many civilians were killed by U.S. airstrikes as had been killed in the 23 years prior.
In setting out an action plan to strengthen the U.S.’s ability to collect, analyze, and respond to civilian harm, the CHMRP-AP and DODI reflected on lessons learned by the military itself about the importance of protecting civilians in warfare. Military leaders echoed this, emphasizing CHMR’s role in enhancing operational effectiveness and preventing mistakes. In a letter to Hegseth, nine military veterans’ organizations stated that CHMR programs “support mission readiness, expand battlespace awareness, preserve strategic advantage, help maintain public trust, and honor the values [service members] swore to defend.”
Answering a poll about CHMR resourcing, every combatant commander opted to maintain the program as much as possible, citing its role in mitigating target misidentification risks and increasing battlespace awareness. Yet the 2025 civilian casualties report provides the latest evidence that the administration has ignored these lessons. As long as U.S. military operations continue without robust learning processes in place, mass casualty events will continue to occur.
