wachiwit - stock.adobe.com
As the war in Ukraine rages into its 16th deadly month, investigators far from the frontlines are trawling through thousands of social media posts and hours of footage in an attempt to hold Russian soldiers and military leaders accountable for war crimes they’ve perpetrated since the illegal invasion began.
Working for Bellingcat and the Global Legal Action Network (GLAN), the investigator’s open source methods – which entail using publicly accessible information (usually social media posts, videos and satellite maps) to explore and geolocate potential crimes – were once a pretty nascent field.
However, it’s becoming an increasingly accepted and acclaimed part of journalism, with major units being launched over the past few years at the BBC, The New York Times and more.
Other groups, such as Forensic Architecture, have been using open source information to investigate human rights abuses.
But now, Bellingcat and GLAN are focused on taking open source investigating to courts, a novel quest that might not only change the way we document war crimes, but also the structure of the globe’s entire legal ecosystem.
The project properly started back in early 2019 during a hackathon event at King’s College London (KCL), which brought together investigative journalists, tech experts and open-source investigation pioneers.
“It was an amazing group of people, and we essentially trialed an earlier version of the methodology we’re now using for Ukraine to look into studying coalition airstrikes in Yemen,” says Dearbhla Minogue, a senior lawyer at GLAN who has been working on the project since its foundation.
The benefits, for Minogue, are obvious. In war crimes investigations, incidents are often being looked into years or even decades after the fact – witnesses can go missing or forget what happened, while physical evidence of the crime has long since disappeared. But social media posts can act as a contemporaneous log of events as they happen, meaning nothing gets lost to the tyranny of time.
Minogue cites how in some of their early investigations into Saudi coalition airstrikes in Yemen, they were able to identify perpetrators based on parts of weapons proven to be at the scene on video, or could prove (contrary to what was being claimed) that air strikes were conducted on civilian sites, based on video footage from just before attacks hit.
The core problem with their aim, though, comes down to something quite simple: belief.
“We shouldn’t forget the first victim of war is always truth,” says Maria Varaki, co-director of the War Crimes Research Group at KCL. “People often don’t want to believe, based on ideology or religion, the outcomes of the International Criminal court.”
That already-present problem is made all the worse when the “explosive cocktail” of technology, as Varaki puts it, gets involved.
In a world of deepfakes, shadowy data harvesting and large language models such as Chat GPT, trust in internet content has plummeted to record lows from the already low levels registered in 2019. If you take evidence people already have a cynical view on and use it in a setting where they’re potentially most likely to oppose it, you can see why some problems would arise.
Another issue comes from the fact that once digital content enters social media, it begins a life divorced from the person who uploaded it.
“One major thing that’s really interesting is how to actually introduce this evidence in court, because the person who filmed the video isn’t there, and that’s usually how a video will get introduced,” says Minogue.
That tendency to have videographers in court for the assessment of their content means there’s usually someone who can take responsibility for and provide context on when, where, how and why the video was made – context you will only ever get a small share of if that content is introduced by a third party who found it online.
Because it’s not had that use case yet, it’s hard to tell how people would react to something as high-profile as a war crimes trial being so digitally focused. But Variaki stresses that its unprecedented nature means the already-high bar for evidence in a war crime trial is all the higher for social media content.
Raising the bar
First, there’s the question of how they select the events they want to analyse. “You have to make sure it’s a case that’s relevant, that is likely to end up in court at some point, and is one where open source investigation primarily based on images and videos actually make a meaningful contribution,” says Nick Waters, the head of Bellingcat’s Justice and Accountability unit.
The exact method they use is shaped by the Berkeley Protocol, a gold standard for open source investigating established by The Berkeley Human Rights Center, which, among other things, established 14 principles to govern how to do thorough, credible, transparent investigations using open source data.
Bellingcat also uses a specialist software called Hunchly that records and tracks everything you do on your computer to ensure there is an audit trail for all the work (in essence to prove how they found the evidence they did, and ensure it wasn’t selectively used). They also work with a group called Mnemonic, which archives and securely stores all the data from its investigations so it is secure from tampering.
Even then, the group is fighting an uphill battle. Computer Weekly has reported before on the often conservative approach to technological change that hampers the legal system’s ability to respond to, say, algorithmic injustice – and in this area it can often be no different.
“Using this kind of evidence is definitely overdue, and there needs to be a proper understanding of it,” says Minogue. “And the levels of understanding are really varied across the legal system.”
Varaki was more sympathetic to the courts’ position (“I would say it’s required to be more cautious”), but the groups working on taking this to court have slowly started to make headway in getting their investigations into courtrooms.
In February 2021, GLAN and Bellingcat managed to stage a mock hearing before a UK Crown Court judge to see what would happen if you tried to introduce open source evidence in court. The judge in that case – who actually went on to join the International Criminal Court – helped build the principles that allowed for the work to be ready to be put into practice – initially planned to be in relation to the ongoing war in Yemen.
While the group struggled to get together funding for the project, Russia invaded Ukraine, and their work saw unprecedented funding and interest, as well as a new European focus – enough to then officially found the Justice and Accountability unit.
The work is complicated by the fact that Bellingcat’s journalistic arm is able to do what its legal arm can’t.
Waters cites Bellingcat’s seminal investigations into the Russian intelligence service’s use of chemical weapons as an example, where they supported public information with phone data on individuals sourced from data brokers. He says that not only does that kind of work raise moral questions over the fact it necessitates engaging with a “criminal economy”, but practically it’s not something that could ever be admissible in court.
Then there are issues around the need for credibility which go beyond standing up individual investigations. Bellingcat, for example, has faced various allegations of working for the CIA or being a government “psy-op” (the latter coming from Elon Musk).
The criticism is largely based on the fact a small chunk of its funding comes from The Atlantic Council, a US think tank with close ties to Western governments, and The National Endowment for Democracy, a non-profit set up by American civil servants to promote democracy in other countries, which is largely funded by the US congress.
This is despite Bellingcat having a long track record of critical reporting on US foreign policy, including directly on US-led airstrikes that killed civilians and the police violence against Black Lives Matter protestors.
Due to the risk of that perception undermining the work, Waters says Bellingcat maintains a strict “firewall” between its journalists’ work and the justice and accountability division, which doesn’t directly report to or take orders from the group’s leadership.
Read more about technology in war
- AI interview: Elke Schwarz, professor of political theory: Elke Schwarz speaks with Computer Weekly about the ethics of military artificial intelligence and the dangers of allowing governments and corporations to push forward without oversight or scrutiny.
- UK, US and Australia jointly trial AI-enabled drone swarm: British, American and Australian military organisations have trialled the use of artificial intelligence (AI) in drones in a collaboration designed to drive their adoption of AI-powered military tools.
- What can security teams learn from a year of cyber warfare?: With the passing of the first anniversary of Russia’s invasion of Ukraine, we reflect on the ongoing cyber war, and ask what security leaders can learn from the past 12 months.
All that demonstrates one thing – the sheer level of work that goes into documenting these crimes. The level of work needed to do all that verification can mean they can end up with 100-page reports on each potential war crime they investigate, according to Waters.
This is a problem, at least when it comes to the Russian invasion of Ukraine, because of the sheer number of war crimes being committed.
Ukrainian prosecutors have alleged that they have collected evidence of some 74,500 war crimes (as of March 2023) committed by Russia since its invasion. Bellingcat’s own civilian harm tracker tool has found open source evidence of just under 1,000 incidents of civilians being harmed during the war over the past year.
Not only is there an obvious arithmetic problem there (the difference between Bellingcat and Ukraine’s figure demonstrates how many war crimes are never documented and posted online), but it also demonstrates an issue with how Bellingcat and GLAN select the cases they do. When you have limited resources and a small team, how do you choose which individual crimes out of thousands to investigate?
“Capacity is the main reason, we had to come up with a set of criteria, and then just make calls on things,” says Minogue. “And there are lots of things that we would like to look at. But we just had to take a decision on what would be the best use of our time.”
War crime trials
In the end, war crime trials are some of the hardest to take to court, not just because of the stakes or burden of proof needed, but just simply getting the accused into a courtroom when the states in question have no desire to turn over soldiers or even leaders to international authorities in the Hague.
Former Serbian president Slobodan Milošević, for example, maybe the most renowned figure to face an international tribunal in the past few decades, only came before the courts years after the Bosnian Civil War ended, once he was forced out of power in an election.
For the Russian-Ukrainian war, it’s likely it would take nothing less than the total downfall of the current Russian government for any of its leadership to ever face charges for what has happened in Ukraine.
That means there’s a chance Bellingcat, GLAN and everyone else looking into war crimes may archive all this evidence only for it to never see the inside of a courtroom.
“It’s something I’m concerned about,” says Waters. “The thing is, if we didn’t do this work, there would be a 0% chance any of it would be used in a prosecution. But if we do this work, there’s still a small chance, and I will take that small chance over zero chance any day.”