Foreign Policy: You started doing this work from your couch. When did you realize that you were actually making an impact?

Eliot Higgins: I started realizing this was a serious thing when I was invited by Tactical Tech, a Berlin-based organization that trains human rights advocates. There were lots of people there from real, serious conflicts. They were coming to me and saying how inspirational my work was to them. I thought, “Wow, if that’s the kind of people I’m inspirational to, I should probably take this a bit more seriously.”

FP: What does the Bellingcat research process look like?

EH: One of the things this technology is about is how much we can automate. We could scan social networks for anyone in a uniform. It’s easy to find camouflage or a shade of green. Then you can start training artificial intelligence to do finer and finer tasks.

[The world’s authoritarians are on the march—and the West helped pave the way, Robert Kagan writes.]

FP: Are militaries and intelligence establishments sometimes shocked at what you’re able to find?

EH: Anecdotally, there have been a few interviews or articles about our work, or they’ve spoken to intelligence people, and they’ve said they’re impressed.

FP: Beyond being flattered, that doesn’t scare you a little bit?

EH: You kind of assume they’re doing all the James Bond stuff—they’ve got rows of computer screens, and everyone’s social media profile is being looked at, that all that stuff’s going on—but, no, it’s not. I’ve been doing a lot of training with the police recently. Even the ones who specialize in open-source investigations tend not to be at the same level that we are.

FP: Is there a specific personality at Bellingcat? People who grew up reading Jane’s Intelligence Review and playing with war toys?

EH: Getting a balance between being obsessive enough and not also crazy is rather difficult.

FP: Can you tell the story of one of the investigations?

EH: We investigated a social media campaign by the Islamic State where their followers in Europe would take photographs holding a piece of paper with the city they’re in and a hashtag. The idea was the Islamic State was everywhere. I saw those photographs popping up on Twitter and thought, “Some of those look like they could be geolocated, but I don’t have time.” So I turned that into a crowdsourced project and had most of the locations in 10 minutes.

FP: When the poisoning of the former Russian spy Sergei Skripal and his daughter happened, did you jump on it when the two suspects appeared on Russian TV?

EH: The Russians may not have realized just how much information was actually out there. In the 2016 Russian-backed coup attempt in Montenegro, an officer from the GRU, the Russian military intelligence agency, was arrested, and he had two IDs: his real one and his fake one. There were items, such as his first name, date of birth, and place of residence, that were the same on both documents.

The theory we were working on was that a Skripal suspect might have done the same. We then used his leaked residency documents, and we had a list of potential names it could be. It wasn’t a massive amount.

The suspects had nearly sequential passport numbers. Also, we found that they had registered their cars at the office of the GRU because it meant when they got pulled over for drunk driving or speeding, the police would look at where they were from and they’d let them go. So now that the Skripal suspects were all potential GRU officers, we had their real names and addresses and identities.

[Gen. Stanley McChrystal explains exactly why Qassem Suleimani is so dangerous.]

FP: Deliberate blurring of the truth seems to be a core part of the Russian information war. If the strategy is to blur truth, then does it matter to have a slam-dunk case in today’s media environment?

EH: There’s also the value of inoculating people against false information. The Russian Defense Ministry was using video of a screenshot from a computer game as evidence that the United States was helping the Islamic State. A few weeks earlier, someone had used the same video to claim it was a U.S. aircraft bombing a convoy. And I’d noticed that, and I said, “No, it’s from a computer game.”

The people who follow me on Twitter are the same people who follow the Russian Defense Ministry. So literally all the replies were people posting that video and saying, “That’s from a computer game.” That’s the only time I’ve seen the Russian Defense Ministry retract a statement because, in a way, people are inoculated against that particular piece of false information.

FP: Is there a way for groups such as yours to expose and push back against deep fakes [computer-generated video or audio that seems real]?

EH: You can’t go to a judge and say, “That video’s fake news.” If you make a fake, maybe you can tweet it, and you’ll get 10,000 retweets. But if you have a video of Barack Obama saying that he regrets not bombing Syria, you want to look at: Where did he say it? Can we find the original video? Why is it not there?

It’s the difference between the impact it has when it’s shared and the impact it has after it’s been verified and used as evidence. There could be a point where they make a deep fake that changes a Russian jet to a U.S. jet. But people are developing tools to look for fake information, so it’s going to start coming down to trusting the sources you use and the people who are sharing information.

This conversation has been condensed and edited for publication. This article originally appeared in the Winter 2019 issue of Foreign Policy magazine.