Last month, NPR released the U.S. military’s internal civilian casualty assessment for the 2019 raid in Idlib, Syria, that led to the death of ISIS leader Abu Bakr al-Baghdadi. In a 14-page review, the U.S. military took evidence provided by NPR and Airwars – the watchdog group I work for that monitors the civilian impact of U.S. and other military actions in conflicts across the globe – and concluded that no civilians were harmed in the raid. This was despite the extensive testimony to NPR of Barakat Barakat, a Syrian man who was travelling home from work with two friends when U.S. troops targeted their van and hit it with airstrikes. Barakat lost an arm and his two friends died. The review ultimately concluded the men demonstrated “hostile intent” by failing to heed warning shots that were fired seconds before the strike.
But the document, which the Pentagon released to NPR after a Freedom of Information Act lawsuit, also revealed something significant – and deeply troubling – about the standards the U.S. government uses to evaluate who it kills in war, and how those metrics may be changing. For the first time, our analysis found that one of the factors the reviewers cited in rejecting the allegation was the lack of metadata on images, including those of the burnt out car Barakat and his friends were in at the time of the strike. If the Defense Department plans to apply this standard going forward, it will be nearly impossible for the U.S. government to accept responsibility when civilians are illegally targeted or killed during military operations.
The Role of Metadata in Civilian Casualty Assessment Reports
In simple terms, metadata is information contained within the original copies of images or videos. Every single image you take contains metadata – allowing you to know the exact time and location it was taken. But most social media companies, including Facebook, Twitter and WhatsApp automatically strip images of metadata for sensible privacy reasons.
Imagine a bomb hits a house on your street, killing a family. As you rush to escape the area you snap a few quick pictures and post them to a WhatsApp group or on Facebook. Those images are then shared online extensively, as is typical in today’s world, becoming evidence of the human toll of the strike. But these images, as they were shared online, would not contain any metadata.
When Airwars, or NPR, or any other organization raise allegations of civilian harm with the U.S. military they send all the evidence available, including images, local press and social media reports, and interviews.
Until now, the U.S. military appears to have never required metadata for images to be considered as evidence.
Using A New Standard?
During the U.S.-led campaign against the Islamic State in Iraq and Syria Airwars sent hundreds of allegations of civilian harm to military reviewers. They would assess the evidence and decide whether they accepted civilians were killed based on factors such as information on the location of the claim in relation to their strikes. The reviewers would sometimes come back requesting additional details through so-called Requests for Information, or RFIs. In our review of 93 RFIs spanning 2016-17 not once was the question of metadata raised.
Working with a series of academics, Airwars is also systematically reviewing 1,300 civilian harm reviews released to The New York Times. This research ultimately aims to provide the most complete understanding of how the United States decides whether those it kills are militants or civilians. But in our review of those 1,300 documents a request for metadata does not seem to appear once.
Yet the civilian casualty assessment report (CCAR) of the al-Baghdadi strike released to NPR contains multiple references to the lack of metadata. On page 9 of the document, the U.S. military analyst questions the veracity of the publicly available information, later noting that “No images included in NPR’s article, on airwars.org or social media/local media sources had metadata for additional time/location analysis.” The lack of metadata is referenced a further three times in the document.
If the U.S. government requires the metadata, it essentially means researchers will have to find the exact person who took the original image, speak to them, and convince them to send the picture directly (assuming they even still have the device and the original copy). In active conflict zones that is nearly impossible.
It remains unclear whether this is a one-off or a significant new trend in U.S. policy. CENTCOM, the U.S. military command covering operations in the Middle East and responsible for processing and assessing civilian harm claims in the region, has not announced any new guidelines. Indeed, an RFI sent to Airwars in 2020 – a year after the Baghdadi raid – made no reference to requiring metadata. The U.S. military later deemed the allegation credible and admitted to having killed one civilian. It comes at a time that the U.S. government is purportedly overhauling its civilian harm processes – with the announcement last year of a major policy initiative called the Civilian Harm Mitigation and Response Action Plan.
In theory these policies should vastly improve how the United States monitors civilian harm from its actions, and make it easier for civilian victims to claim compensation when they are harmed. Yet if metadata is a new standard by which reviews are conducted, it will become almost impossible for the United States to accept that its actions killed civilians.