Latest / Stage Talks with Bellingcat / "Using the Open Source Munitions Portal" with The OSMP Team
Transcript
- 0:00You're listening to a stage talk titled, Using
- 0:02the Open Source Munitions Portal. This week we
- 0:06spoke to Joe Dyke and Joe Emmett from Airwalls
- 0:09and Nick R. Jensen -Jones from Armant Research
- 0:12Services. Together they work on the Open Source
- 0:15Munitions Portal, a database of different munition
- 0:18types for open source research analysis. We spoke
- 0:22on ways to use the tool, their latest updates
- 0:24and why resources like theirs are important.
- 0:27This talk was hosted by me, Charlotte Ma, on
- 0:30Thursday the 13th of March 2025 in the Bell and
- 0:34Cactus Discord server. Today we're talking about
- 0:55weaponry, specifically munitions identification.
- 0:58I'm joined on stage by the team behind the open
- 1:00source munitions portal. The portal is a free
- 1:03to use tool that aims to provide a key resource
- 1:05to democratize information about conflict and
- 1:08open up access to understanding munitions use.
- 1:11Bellingcat and our wider volunteer team use it
- 1:14frequently. The OSMP offers nearly a thousand
- 1:18reference images of a variety of munitions from
- 1:20active conflict zones, each reviewed and assessed
- 1:23by at least two munition experts. Key information
- 1:26is provided about each image, including reported
- 1:29location, tentative model of munition, and where
- 1:32relevant analyst notes written by experts. You
- 1:34can also find guides to munition types, explainer
- 1:37videos, and a glossary of key terms on site.
- 1:40Basically, they have everything you would need
- 1:42to know about munition types on their website.
- 1:45Recently, the team launched the first of a series
- 1:47of articles using 3D interactive models to explain
- 1:51how to identify some of the most widely used
- 1:53munitions, which we will hopefully dive into
- 1:55today. The site has agreements with a number
- 1:58of other research organizations, including ourselves
- 2:01at Bellingcat, Human Rights Watch and Myanmar
- 2:04Witness, allowing access and re -publication
- 2:06of their archive of munitions images. The team
- 2:09next to me is Joe Deit, director of programs
- 2:11at Air Wars and a co -founder of the OSMP. Nikar
- 2:15Jensen -Jones is also a co -founder and an arms
- 2:17and munitions specialist and a director of armament
- 2:20and research services. And Joe Emmert, a researcher
- 2:24for Air Wars with the ears of open source intelligence
- 2:26experience, focusing on conflict zones and munitions
- 2:29use in the Middle East and Ukraine. If you'd
- 2:32like to ask a question, please do so in the chat
- 2:35as we talk. And please note, as I just mentioned
- 2:38within your question, if you do not want me to
- 2:40read your username out, because this is being
- 2:42audio recorded and will be posted on podcast
- 2:45platforms following this session. Okay. I'll
- 2:49be back with the questions in a second, particularly
- 2:51those that you ask in the chat. But first, let
- 2:54me pass the mic to the OSMP team so we can learn
- 2:57a little bit more about this fantastic resource.
- 3:01Guys, take it away. Okay. Well, yeah, thank you
- 3:04first of all for coming. It's really interesting.
- 3:07We're going to try and keep our talks like you
- 3:09asked to 20, 25 minutes. So we do, we will obviously
- 3:14be talking and explaining the site, but we know
- 3:16it's a podcast as well. So we'll try and keep
- 3:18it broad enough that If you're not looking at
- 3:21what we're looking at, you can still kind of
- 3:22understand. I'm going to talk briefly about where
- 3:26the idea for the OSMP came from, and particularly
- 3:29the problem it was trying to solve. Nick's then
- 3:32going to talk briefly about how he and the Ares
- 3:36team designed the kind of verification and...
- 3:41methodology to a certain extent. And Joe then
- 3:43we'll talk about what's actually on the site
- 3:45and some of the new developments, including,
- 3:47as you mentioned, 3D models, et cetera. So as
- 3:51a kind of brief background of where this kind
- 3:53of came from, I worked as a journalist in the
- 3:56Middle East for a decade. My last job being chief
- 3:59correspondent for the West Bank in Gaza with
- 4:01AFP, which is a big global news agency. I was
- 4:04more of a kind of traditional, no, non -OUSINT
- 4:07journalist. working on conflicts, writing about
- 4:10conflicts, but I never really had the time and
- 4:13accessible resources to learn properly about
- 4:15munitions. And I know that was a thing from my
- 4:18peer group, people that, you know, other journalists
- 4:20working in their world, that we all kind of felt
- 4:22like it was something we really wanted to learn
- 4:24about, but didn't really have the time or the
- 4:27expertise. And then in 2021, I joined Air Wars.
- 4:32and set up the investigations unit. And we began
- 4:34using open source techniques to investigate civilians
- 4:38being killed in conflicts across the globe. For
- 4:42those of you aren't aware of what we do, Air
- 4:44Wars is the world's leading open source documenter
- 4:47of civilian harm, assessing civilian harm claims
- 4:51in countries across the conflicts across the
- 4:52world, including Syria, Yemen, Ukraine and Gaza.
- 4:57We were set up actually the same year as Bellingcat
- 4:59in 2014, and actually a lot of the techniques
- 5:02that Bellingcat has pioneered, geolocation, et
- 5:05cetera, we have built into our systems and the
- 5:08two organizations have overlapped many years.
- 5:12But then when I started doing investigations
- 5:14at Airwalls, essentially I kind of realized after
- 5:17doing a few that I was doing the same process
- 5:20every single time as an OSINT process, where
- 5:23I would kind of gather all images of munition
- 5:27remnants, craters, et cetera. And then I would
- 5:30send them to one of the five or six munition
- 5:33experts that I knew to review. And kind of over
- 5:37time, I kind of realized this was both inefficient
- 5:40and pretty undemocratic. It was inefficient in
- 5:43that every time I was asking a favor from the
- 5:46same few experts, sometimes they would come back
- 5:49and say, yep, that's a Russian artillery shell
- 5:51model number, blah, blah, blah. And sometimes
- 5:54they'd come back and say, yeah, you've just sent
- 5:55me some metal. That doesn't really tell me anything.
- 5:59Nick maybe can speak a little bit more to that,
- 6:01but you had a few experts being inundated requests,
- 6:05often multiple journalists sending them the same
- 6:07images about the same incident asking for verification.
- 6:11And it was also kind of like inefficient in that
- 6:14I wasn't actually learning or getting better.
- 6:16I was just doing the same thing and kind of outsourcing
- 6:19that every time. And it was kind of undemocratic
- 6:22because I work for Air Wars and so we have contacts
- 6:25with munitions experts and those kind of things.
- 6:27The vast majority of journalists do not have
- 6:30that. And that includes journalists in conflict
- 6:32zones, journalists in smaller news organizations,
- 6:37people who are kind of experts in OSINT investigations.
- 6:40So they don't have that network of people that
- 6:44they can go and check information. And so there
- 6:47really was no resource available to fill that
- 6:50gap. And so a colleague of mine and I basically
- 6:53said, wouldn't it be great to set up an accessible
- 6:56library of munition images that had been reviewed
- 6:59by actual experts? And so that was kind of the
- 7:03idea. And then I called up Nick, who will speak
- 7:06in a second, who I'd known for a little while
- 7:08by that stage. And he is one of the world's leading
- 7:12experts in munitions and was one of the people
- 7:14that I would call up and ask if they could do
- 7:17me a favor. And it turned out he'd kind of been
- 7:20having the same idea for a number of years at
- 7:22the other end of the telescope. Aries have trained
- 7:26hundreds of journalists and researchers and others
- 7:29in munitions, but at the end of the courses,
- 7:31et cetera, they don't really have easily accessible
- 7:34material to direct them to. Maybe six months
- 7:36down the line, a journalist would be like, oh,
- 7:38I did that really interesting course, but now
- 7:41I've forgotten it all and now I kind of want
- 7:43to come back to that information. And there was
- 7:45nothing in the public. domain that's like easy
- 7:48and accessible. Images also, you know, referent
- 7:52images tended to be munitions in their entirety,
- 7:55not how people, investigators, journalists, et
- 7:58cetera, find images of munitions in the real
- 8:01world, you know, once they have exploded, you
- 8:04know, a particular fragment, et cetera. And even
- 8:08as like practitioners, experts like Nick, who
- 8:11can maybe talk a little bit more, but the of
- 8:14reference material meant that like going and
- 8:17saying, okay, I want to check my assumptions.
- 8:18I'm pretty sure that this is this, but I want
- 8:19to just see five other pictures of that just
- 8:22to verify was sometimes unnecessarily time consuming.
- 8:26So for that, Nick and I basically kind of came
- 8:29up with the idea for the tool, which we described
- 8:32as a tool for journalists, human rights researchers
- 8:35and lawyers who often receive little training
- 8:38in munitions and have few accessible resources.
- 8:44And the aim was not and is still not to turn
- 8:47anybody into a specialist. Munitions, as I've
- 8:50learned over the last few years, are incredibly
- 8:53complicated. Every time I think I've understood
- 8:55a munition, Nick or somebody will say, well,
- 8:58actually, there's an additional new variant.
- 9:01And you always have to be aware of false assumptions
- 9:04and matches. And we do not say that the tool
- 9:07can make you a munitions expert. But what we
- 9:10aim to do is to raise everybody's level up a
- 9:14little bit of understanding, a little bit, and
- 9:17to build an archive that would allow you to learn
- 9:19from investigations that have been done over
- 9:21the last 20, 30 years in an easy and accessible
- 9:25fashion. And so we applied and got a grant from
- 9:29the Google News Initiative and now from the European
- 9:32Media and Information Fund. And we built a prototype,
- 9:35which we launched in 2023. a month or so after
- 9:39the war in Gaza had begun. That was a very kind
- 9:43of janky prototype. So we then spent five months
- 9:48getting in -depth feedback from journalists,
- 9:51lawyers, human rights researchers, and others,
- 9:53specifically about how the tool could be more
- 9:55useful to them, what functionalities that we
- 9:57didn't have, et cetera. built a whole new version.
- 10:02It's not actually the same websites that we started
- 10:04from scratch. And that was eventually launched
- 10:06in June last year. And we see this tool and its
- 10:10evolution is kind of pushed towards more accurate
- 10:12and responsible reporting on conflict. And we
- 10:16at Air Wars are kind of geared around this mission
- 10:18to focus attention on the full impact of conflict
- 10:24on, particularly on civilians. And yeah, that
- 10:28is... We, the, I'm going to hand over to Nick
- 10:31here, but kind of my last point would be that
- 10:33we would, we have, we still see this as like
- 10:36a collective tool. So like, we do genuinely want
- 10:39feedback because the things that aren't there
- 10:41that could be really, really useful, uh, to people,
- 10:45uh, where we also want to know those. So yeah,
- 10:47that's, um, that's me and Nick, maybe I'll hand
- 10:49over to you. Yeah, thanks Joe. Um, right. Uh,
- 10:56I think probably, uh, some of you are familiar
- 10:57with. with ARIES, but for those of you who aren't
- 11:00ARIES, Armament Research Services, we call ourselves
- 11:02a specialized technical intelligence consultancy.
- 11:06And essentially what we do is provide detailed
- 11:07answers on questions about arms and munitions.
- 11:09So everything from proliferation questions, how
- 11:11did that get there? Where might it end up next?
- 11:13To very technical, what we'd call pins and springs
- 11:15questions. How does it function? You know, one
- 11:18of the key components we can look at. So we came
- 11:21on board in part when Joe and I spoke the first
- 11:25time because we had, as Joe said, received so
- 11:27many requests for the same sort of answers from
- 11:30various customers, if you like, or mostly unpaying
- 11:34customers that really needed sort of fairly basic
- 11:36information from our perspective, but needed
- 11:39it to be verifiable and needed to be confident
- 11:43in the answers we were giving them. So you can
- 11:45imagine that a group of technical specialists
- 11:47with really niche skill sets and Joe is very
- 11:50kind in what he said about me. I would maybe
- 11:51push back a little and say my greatest strength
- 11:53is that I know a lot of really brilliant experts
- 11:55in very niche fields. We have a guy who, the
- 11:58example I love to give is we had a guy once upon
- 12:00a time who specialized in East German ammunition
- 12:03packaging from the 1970s. That was his focus.
- 12:06And he would look at the fonts and the kerning
- 12:08and look at when they changed the spacing between
- 12:10the R and the E and this sort of thing. You think,
- 12:12oh, okay. So having access to those people, they're
- 12:15often coming in at a 10, if you like, at full
- 12:17bore. nails, pins, springs, bolts, whatever it
- 12:20is, they're really down the details. And a journalist
- 12:23or an NGO user, or even an independent researcher
- 12:26might need a five or a six, right? And what's
- 12:28out there on the internet might be a one or a
- 12:30two, or it might come across as a five or six,
- 12:32but be completely unverifiable. So the OSMP in
- 12:35many ways is about confidence. And what Aries
- 12:38does, I think what we bring to the table for
- 12:40this is giving the user the confidence in what
- 12:44they're reading. So you've got to have the really
- 12:46high level as you say Charlie, niche expert sources,
- 12:49but you also need to have some of those generalist
- 12:52introductory sources. And ideally you need to
- 12:54have something in between. And that was the bit
- 12:56that Joe and I really felt was missing. We thought
- 12:58you could get some introductory training from
- 13:00people like us and others. You get some really
- 13:02competent technical specialty work from, again,
- 13:05us and others, but you really had a hard time
- 13:08if you wanted to get on -ramp into some of the
- 13:11sort of medium sized questions about arms and
- 13:13munitions. So Aries in contributing to the OSMP,
- 13:17a large part of what we do is the verification
- 13:19process. So we contribute in some ways to identifying
- 13:22important information that might go into the
- 13:25tool, but really what we're doing is verification.
- 13:28So we're checking those images, making sure that
- 13:30they're of sufficient quality, that the object,
- 13:32the munition can be classified, and then formally
- 13:35assessing them through a process that sees it
- 13:38pass through at least the hands of two different
- 13:40ARIES reviewers. So sometimes it's three. If
- 13:42something's really specialized, it might be four.
- 13:44But generally speaking, if you see something
- 13:46on the OSMP, Two ARIES technical specialists
- 13:48have looked at it and they've made a tentative
- 13:51identification wherever possible. We make a distinction
- 13:55between identifying and classifying. And those
- 13:57are two kinds of operations that are really useful
- 13:59in arms and munitions investigations. Sometimes
- 14:02people need to look at weapons at a much more
- 14:04granular level. They need to know for attribution
- 14:06purposes, for example, the make and model and
- 14:08the manufacturer, what country it came from.
- 14:11For forensic purposes, you need to go even more
- 14:13granular and it might need to match a projectile
- 14:15to a... gun barrel, you might need to match a
- 14:17powder residue to a rocket launcher blast, for
- 14:19example. But other times you need to look at
- 14:22things at a much more macro level and you need
- 14:23to understand, you know, what types of weapons
- 14:26they are in broad categories or how many air
- 14:28delivered munitions were used in this conflict,
- 14:29how many guided munitions, whatever it might
- 14:31be. So OSMP tries to sort of give people both
- 14:35sets of data where possible. But as Joe said,
- 14:37we're not pretending that we can put a tool on
- 14:40the internet that makes you into a specialist
- 14:41with decades of experience. Instead, we're helping
- 14:44people walk themselves down that path and then
- 14:46when they can make an informed decision about
- 14:48what kind of information they might need next.
- 14:50So if you know you only need that sort of macro
- 14:53level classification type information, then maybe
- 14:55you don't need to go and pay a lot of money to
- 14:57consult someone like me to tell you something
- 14:59you don't need to know. On the other hand, if
- 15:01you go down the path of understanding the weapons
- 15:03in a particular investigation that you're interested
- 15:05in, and those arms and munitions, turn out to
- 15:08be really complicated or you need to interpret
- 15:11them in a way that requires specialist expertise,
- 15:12then you at least have a starting point to go
- 15:14to someone and have a conversation with them
- 15:16and explain what you need and what you've already
- 15:17done. So that was the gap we thought needed filling.
- 15:21The verification process, like I say, is I think
- 15:23where we put our best foot forward so that we
- 15:25give people confidence. When you look at the
- 15:26OSMP, you're confident that, you know, there's
- 15:29a, there's a small likelihood that there's an
- 15:31error in there. We're fairly confident. You know,
- 15:33obviously it comes with all the usual caveats.
- 15:35We're doing this from open source material. We're
- 15:37doing this from a distance. In most cases, some
- 15:39of these we've, you know, there might've been
- 15:41a physical interaction, but in almost all cases,
- 15:43these are done from a distance from test -based
- 15:45research. So there are all the typical caveats
- 15:48for identification, but at least the level of
- 15:50confidence is hopefully higher for the average
- 15:53user than it would be if they'd stumbled across
- 15:55the photos themselves. I think that's probably
- 15:59the gist of it in the time that we have, but
- 16:01Joe can talk a little bit about the process that
- 16:05each entry goes through. Cool. Okay. Yeah. Hi
- 16:11everyone. I'm the other Joe. I joined Airwars
- 16:16in March of last year and essentially my job
- 16:20is more at the start of the documentation process.
- 16:23So I will be the one sifting through social media,
- 16:27so telegram pages, Twitter accounts, Facebook
- 16:30groups to essentially find the actual munition
- 16:34remnants and then add them to the portal with
- 16:36the initial tags before they go to the technical
- 16:40specialists from Ares. I kind of want to start
- 16:44by taking you through the evolution of the OSMP.
- 16:47So where we've come from and where we are now.
- 16:51Essentially, as Joe said, we initially launched
- 16:53in beta in November of 2023, and then we relaunched
- 16:59last year in June 2024 with a brand new design.
- 17:03And throughout this process, we've sort of been
- 17:05aiming to add somewhere around 50 new munition
- 17:07images each month. And that's allowed us to expand
- 17:10from around 230 to now about 860 munition images.
- 17:16And within these munitions, they cover around
- 17:18200. different models and types from around two
- 17:23dozen different locations, regions, countries
- 17:25across the world. And throughout this, we've
- 17:29also been trying to expand the categories of
- 17:32munitions. So not just air delivered bombs or
- 17:35rockets or missiles. We've now been adding tank
- 17:37gun projectiles, our recoilless rifle projectiles,
- 17:41and most recently drones or UAVs such as the
- 17:44Iranian shayad 131 and 136 or the Israeli harap
- 17:48that was used by Azerbaijan quite a lot in Nagorno
- 17:51-Karabakh. We've also recently begun to sign
- 17:56agreements with certain organizations, one of
- 17:59them being Bellingcat as well as Human Rights
- 18:02Watch and Myanmar Witness to essentially allow
- 18:05us to tap into their world -leading research
- 18:10and investigative practices so we can use their
- 18:14images from the field. We can add them to the
- 18:16portal, add tags and ID them. And a good example
- 18:20would be one from Bellingcat. Look at the research,
- 18:25different research organizations on the portal.
- 18:28And essentially when you click on each of the
- 18:30images, it will allow you to look at the original
- 18:33source, as well as an external reference link
- 18:36where you can go to the organization's page,
- 18:39talking about the munition. Another The feature
- 18:43we've added to sort of aid understanding of munitions
- 18:45is what we refer to as the Analyst Note feature.
- 18:49So that's when we finish reviewing a munition
- 18:52and Ares technical staff will add information
- 18:56to provide basically greater detail surrounding
- 18:58a unique munition capability. So for one example,
- 19:02might be a GBU -39, one of these ones we've got
- 19:04from Gaza. And Aries added that this image shows
- 19:09various munition remnants, including a fuse well
- 19:12and two nose cone fragments from GBU -39 small
- 19:15diameter bombs. The presence of two different
- 19:18nose cones indicates that these remnants are
- 19:20from at least two distinct munitions. So not
- 19:22only can we show you what these pieces of metal
- 19:25essentially are, but we can also say something
- 19:28that you may miss is that we've actually got
- 19:30two nose cones. So it suggests in this instance
- 19:33that there might be two GBU -39. air -delivered
- 19:36bombs that hit this location. Talking about GB39s,
- 19:41they were widely used in Gaza, which brings me
- 19:45on to my next point, which is how we actually
- 19:47not just document munitions, but try to link
- 19:50them to incidents of civilian harm, that we don't
- 19:53overlook the actual impact of the munitions in
- 19:56conflict zones. Through this process, we essentially
- 20:00usually work with members from the Air Wars team.
- 20:03to match munition debris found in a particular
- 20:06location to a specific civilian casualty incident
- 20:09that's being investigated. It can also match
- 20:12images obviously we've found from Human Rights
- 20:14Watch and other organisations to their work documenting
- 20:17the munitions as well. And we'll always add an
- 20:21external resource link to these munitions so
- 20:25you can look at the original work essentially
- 20:27as well. The reason behind adding this is that
- 20:34we think it provides important context, not only
- 20:37of what the munition is, but most likely for
- 20:40people who are investigating say a potential
- 20:42war crime, you can look and see who actually
- 20:45may have deployed the munition, so who has that
- 20:47in their inventory, what platform was likely
- 20:49used. So is it an artillery piece, a tank, a
- 20:52fighter aircraft, something like that. And in
- 20:54some instances where we have markings or data
- 20:56panels on the fragments, we can actually look
- 21:00at who potentially made that, who the weapons
- 21:03manufacturer of that specific part of the munition
- 21:05was as well. And we, essentially in doing this,
- 21:10we're trying to help the Samoan community gain
- 21:13an understanding of what the telltale signs of
- 21:16the aftermath of a strike may look like for GB
- 21:1939 or Mark 84 and what to look for in the future.
- 21:24But I think as Joe mentioned, you've always still
- 21:26got to be very cautious and look around for any
- 21:30false positives in this. In sort of recent months,
- 21:35we've also moved beyond just documenting munition
- 21:38images to actually helping you visualize the
- 21:41munitions. So we've been essentially creating
- 21:463D interactive models of some of the most widely
- 21:50used munitions in conflict zones. which a lot
- 21:53of you may have seen with the GBU -39. So essentially
- 21:56it will give you a 3D model with extra information
- 21:58around it. And you can click on the model and
- 22:01look at the model from various angles, as well
- 22:04as learn about the different parts and match
- 22:05them directly to images in the OSMP to corresponding
- 22:08parts. And in each of these parts, images linked
- 22:12to the certain sections, you can click on them
- 22:14and you can look at the original image, where
- 22:15the source came from, just to get some extra
- 22:17context around it. And we thought the GBU -39
- 22:21was a pretty important starting point as it was
- 22:24one of the ones that the New York Times called
- 22:25Israel's weapon of choice in Gaza. We're also
- 22:28looking at more widespread munitions that are
- 22:31going to be used in the future as well. I'm working
- 22:33on some other 3D models and different iterations
- 22:36of how we can basically mess around and change
- 22:38with the way in which the model is visualized
- 22:41and maybe help you compare different munitions
- 22:43side by side as well. Lastly, I kind of want
- 22:46to talk about how the OSMP has become more reactive
- 22:50due to the nature of the conflicts we document
- 22:53munitions use in. Important incidents occur on
- 22:59short notice in a short period of time and they
- 23:01often generate large amounts of public interest.
- 23:04And so where there is this particular interest
- 23:06or potentially misinformation, we're going to
- 23:09try and respond as accurately and quickly as
- 23:11possible, essentially to provide as accurate
- 23:15as possible information about the munitions used
- 23:18in certain instances. So for example, after the
- 23:21strike on Hassan Nasrallah, the Secretary General
- 23:24of Hezbollah in September of last year by the
- 23:27Israeli Air Force, the Israelis actually posted
- 23:30on the Israeli Air Force Twitter account an image
- 23:33of one of the F -15s involved in the strike.
- 23:36And there was a lot of reporting about bunker
- 23:37busters being used. And we were actually one
- 23:40of the first to be able to identify that the
- 23:42munitions hanging off The aircraft were what
- 23:45we, what were found to be BLU 109 JDAMs, which
- 23:49are essentially bunker buster and what is commonly
- 23:52referred to in the media as bunker busters, confirming
- 23:55that. Another example might be the recent strike
- 23:58on the Chernobyl new safe confinement structure.
- 24:03And there was a reported drone strike on that.
- 24:06And then there was a lot of back and forth claims
- 24:08about who had done it. Russia said that it wasn't,
- 24:12that they weren't essentially behind the strike
- 24:14and often even accused Ukraine potentially of
- 24:16being behind it. The issue with that is that
- 24:19there were images of debris and remnants from
- 24:22the scene. And we matched one of these images
- 24:25to a video and some other images from one of
- 24:27the gantries, the walkways at the upper levels
- 24:29of the New South confinement, which showed what
- 24:33we found to be an MD550 motor, which is the engine
- 24:37essentially of Iranian and Russian made Russian
- 24:40variant Shahed 136 UAVs or kamikaze drones as
- 24:45they're often referred to, which would suggest
- 24:47that Russia was most like the one behind the
- 24:49strike and not Ukraine. And yeah, we've essentially
- 24:53tried to expand that and remain as reactive as
- 24:55possible and responding to it. So we've done
- 24:58other ones when there was a storm shadow strike
- 25:00in Kursk. We were able to identify some elimination
- 25:03remnants and confirm that a British made air
- 25:06-launched cruise missile was used inside Russia
- 25:08for the first time by Ukrainian forces. So yeah,
- 25:11that's kind of how the OSMP came to be and how
- 25:15it's moved forward. And maybe in a little bit,
- 25:17we can talk about the actual future of where
- 25:20we think it might be going. But yeah, apart from
- 25:22that, we'll be happy to take some questions.
- 25:25Amazing. Thank you so much for doing a little
- 25:27tag team there and really embedding us in the
- 25:31growth of the OSMP. OSMP. It's really interesting
- 25:35to hear. Keep your questions coming in. I've
- 25:37seen a few pop in already. I wanted to start
- 25:40just by asking though, because this is something
- 25:42that often comes up when you're looking at building
- 25:45a tool or building a resource, particularly for
- 25:48open source research where you find a gap, is
- 25:52the difficulty in getting the resources and the
- 25:55time to keep it maintained and to keep it up
- 25:58to date with the flood of information that comes
- 26:01in. When setting up the OSMP, did you find it
- 26:05at first difficult to get time investment for
- 26:08documentation? How did you learn to balance resources?
- 26:13For example, Bellingcat has just introduced toolkit
- 26:15maintainers to keep our toolkit up to date and
- 26:19they are doing an incredible job. Before it was
- 26:21very, very difficult for us to keep up with the
- 26:23pace of different tools being launched and to
- 26:26make sure that our toolkit was representative
- 26:29of the latest open source tools out there. So
- 26:33how has the OSMP team been able to manage that
- 26:37process? Yeah, it's an interesting question.
- 26:40I think when we first set out, we very much said
- 26:44that this was about quantity, not quality. And
- 26:49I think, uh, Nick, uh, and so I think Joe mentioned
- 26:52that we had about 50 images. Now that's not actually,
- 26:55you know, by OSIN standards, that doesn't feel
- 26:58like very much, but they each go through quite
- 26:59a very rigorous process and it takes 50 images
- 27:03per month, you know, is actually a lot of work.
- 27:05And so it's about selecting the images that we
- 27:08think are the most, uh, useful. Yeah. I mean,
- 27:11the Bellingcat toolkit is such a useful, uh,
- 27:14thing, I guess what, what We're doing, we understand
- 27:18that like the archive has cumulative effects.
- 27:21If you know what I mean, that the bigger it gets,
- 27:23um, over time and you had the same methodology
- 27:26over time, it becomes more useful and more, more
- 27:30useful. And, but we, we kind of also as, and
- 27:33then, uh, you know, very much you'll notice there's
- 27:37lots of things that aren't on there right now
- 27:39that we, we were like, okay, let's start with.
- 27:43We initially started with full munition categories.
- 27:46We've now added tank projectiles and recoilless
- 27:50weapons, and then some drones. But yeah, there's
- 27:53a lot of stuff. Mines is not on there. There's
- 27:56also things that are not. So we're trying to
- 27:59do it well. And I think Joe E always says we
- 28:04need a minimum of 10 images of a particular munition.
- 28:10for it to be useful as a reference, right? Cause
- 28:13you kind of want to see it from the same thing
- 28:15from multiple different angles, et cetera. And
- 28:18so, yeah, I think we're not trying to kind of
- 28:22keep up in that sense. And we're not in the same
- 28:24way that Bellingcat is a, like a hub of all of
- 28:28these methodology. Like what we're trying to
- 28:30do is like build kind of slightly methodically
- 28:32a really reliable and in -depth archive, I think.
- 28:37Yeah, exactly. That's amazing. We also spoke
- 28:41previous to when setting up this call about the
- 28:44fact that you also allow people to submit images
- 28:47to the site for verification. Can you talk a
- 28:50little bit about that submission process, what
- 28:53you're looking for and if there's any caveats
- 28:56to content that you can upload for verification?
- 29:02Yeah, sure. I'll take this one. In terms of It's
- 29:08open to anyone to submit an image. We do say
- 29:11generally they should be a decent quality image.
- 29:14So it should be fairly up close to the munition
- 29:16or the munition remnant should be visible. And
- 29:21ultimately you can go to the website and you
- 29:25can, anyone can upload them so long as they have
- 29:27a corresponding source image as well that we
- 29:30can add and we can reference because part of
- 29:32our sort of verification process, which I was
- 29:35going to get into is that We try and look for
- 29:38the most reliable sources that we can take munitions
- 29:43from and get as close to the original source
- 29:45as possible. And then we put it through TinEye
- 29:50and Google Lens. We reverse image search it to
- 29:52essentially make sure that we're getting as close
- 29:54to the original as possible. And we're checking
- 29:58the veracity essentially of the image and making
- 30:01sure that it's as accurate as possible before
- 30:03we even start the process of going through the
- 30:07methodology behind classifying it. But it is
- 30:11open to anyone. Anyone can submit. It sends it
- 30:14directly to us. We get an email each time and
- 30:16then we can review it and potentially upload
- 30:19it. That being said, you do have quite a large
- 30:23PSA on your site about not approaching munitions
- 30:27under any circumstances. You mentioned that you
- 30:31get quite a lot of imagery from organizations
- 30:33like Bellingcat and things like that. Where are
- 30:35most of your images coming from? How do you make
- 30:39sure that people aren't taking unnecessary risks
- 30:42to help document munitions on the ground, for
- 30:45example? Well, I can answer the first bit and
- 30:48then maybe Nick is best equipped to answer the
- 30:51second, the last part of that. Essentially, most
- 30:54of our imagery is coming... maybe predominantly
- 30:59from social media. So different accounts, we
- 31:02search in Arabic language or Facebook groups.
- 31:05We often save them and archive them so we can
- 31:07go back because for example, the Palestinian
- 31:10police force in Gaza have an explosive ordnance
- 31:12disposal team and we're often looking and scouring
- 31:15through that. But it's a whole range of sources.
- 31:20I wouldn't say we try and balance it out, but
- 31:23it's predominantly social media. And then we're
- 31:25also in or signing more and more potential agreements
- 31:29with other organizations to use their imagery
- 31:32that they've actually got people in the field
- 31:34who are going there and taking photographs of
- 31:37the munition. Very often in the past, we have
- 31:40some images from, I think, Human Rights Watch
- 31:43of US Atacans, which is a ballistic missile with
- 31:47cluster munitions from Iraq in, I think, 2000.
- 31:50three or four. So a lot of our images date quite
- 31:54far back as well, especially when we take them
- 31:56from other organisations. But we're trying to
- 32:00essentially get a wide range of sources to do
- 32:03that. In terms of approaching munitions, I think
- 32:05Nick maybe could talk a bit more about the dangers
- 32:08of that. Yeah, sure. I mean, I don't think there's
- 32:11too much that needs to be said. You know, I'll
- 32:13give the obligatory safety warning, which we
- 32:14have on the page, which is don't do it unless
- 32:16you're unless you're trained to do so. As we
- 32:19partner with different organisations, Obviously
- 32:22we want to have confidence that they're not putting
- 32:24their employees or their sources in positions
- 32:26of risk, but ultimately you can only go so far
- 32:28down the chain. So I think the remedy to this
- 32:32potential threat to sources or third order effects
- 32:35that we might not be able to visualize is just
- 32:37promoting good safety practices around arms and
- 32:39munitions. So we use the arms acronym, which
- 32:42is avoid the area, record all the information,
- 32:45mark the area from a safe distance to warn other
- 32:47people. And then seek assistance from the relevant
- 32:49authorities, whoever that might be in a place.
- 32:51So at the simplest form, that just means if you
- 32:54see a munition and you happen to be in the field,
- 32:55you stay away from it. You talk to someone competent
- 32:57and it's really that simple. Very sound advice.
- 33:02And Mike, the spelling expert asked earlier,
- 33:04would you ever look at creating a similar database
- 33:07for full weapons systems, effectively having
- 33:09someone watch a military parade in Russia and
- 33:13Uzbekistan and identifying the vehicle scene,
- 33:15for example? Is that something you've ever thought
- 33:18about? Is that kind of resource out there that
- 33:21you're aware of? Joe, do you want me to tackle
- 33:23this or you want to chime in? We would have thought
- 33:28about that. Yes. We love the idea. I think, and
- 33:32Joe probably has some words to say, but in general
- 33:35terms, there's a lot of things we could do with
- 33:37this platform. We have a big document that we
- 33:41all are very excited about. Of course, it's all
- 33:43about time and money at the end of the day. how
- 33:46we prioritize those different avenues for future
- 33:49development, I think is an ongoing discussion.
- 33:50But as Joe the Younger mentioned, we've already
- 33:52added recoilless projectiles, for example, and
- 33:56we're adding new types of munitions that we're
- 33:57looking at and we're adding more detail on those
- 33:59that are already in there. Going forward, yeah,
- 34:02I mean, in a perfect world, I'd love to see us
- 34:04tackle all arms and munitions and then maybe
- 34:06even look at platforms. Yeah, just really to
- 34:11reiterate, yeah, like it was kind of trying to
- 34:15bite off what we can chew relatively slowly in
- 34:20history, in the sense of we'd rather feel like
- 34:25we built a really strong database of this. of
- 34:28specifically the ammunition types that we're
- 34:30looking at. And then, but yeah, like Nick and
- 34:33I have talked about small arms, there's lots
- 34:35and lots of worlds in which this would be incredibly
- 34:38useful. And the main thing that we did with the
- 34:42second version of the portal was we made it kind
- 34:45of fully scalable, like the structure is scalable.
- 34:48And so that was actually uh, some of the data
- 34:51stuff that, but like it involves some of like,
- 34:53um, stripping back some of the things that we
- 34:55had coded for previously and making it a bit
- 34:57less so that like, yeah, if at some point we
- 34:59wanted to add in mines or we want to add in small
- 35:01arms, yeah, all of that is doable within the
- 35:04portal. And like, um, I think that would be a
- 35:07really useful potential, uh, thing that, you
- 35:11know, over time would be of use to people. There
- 35:16you go, Mike. There's appetite for it. Uh, That
- 35:19being said, while we're talking about resources,
- 35:22we mentioned that people can submit. Is there
- 35:25an opportunity for people to apply or offer to
- 35:29help with the verification process? Is that kind
- 35:31of lopped away to ARIES or is Air Wars volunteers
- 35:36also helping with the verification process? Is
- 35:39there any way that open source researchers can
- 35:42apply to be part of the verification process
- 35:45at OSMP? Might be another one for both of us,
- 35:47but I think I'd say, you know, not to be too
- 35:50proprietary because the point of this, of course,
- 35:51is, as Joe said, to democratize information and
- 35:53get some of this out there, not behind a paywall
- 35:55where it often is. I think the real value in
- 35:58what we're doing lies in the confidence we give
- 36:00users. I want people to be able to look at ammunition
- 36:02and know that they have confidence in how that's
- 36:04gone through the process. So certainly there
- 36:07are avenues for people to join Ares and Air Wars.
- 36:11And we're always looking for more talented people
- 36:13with a specific background and skill set. in
- 36:16identifying arms and munitions in our case, and
- 36:18in other kinds of open source research as well
- 36:20in Airwar's case. But equally, I think, you know,
- 36:23it's important that if people, you know, attach
- 36:27to this project or end up working on the OSMP,
- 36:29they've come from a really strong background
- 36:30that gives confidence to the user base. I think
- 36:32that's what's more important than the volume.
- 36:34A lot of the tools, I had a conversation with
- 36:37this, with a... open source intelligence specialist
- 36:41about this not too longer. And he was talking
- 36:44to me about the sort of different waves of OSINT
- 36:46tools that have come out. And in wave one, a
- 36:49lot of it was all about capturing data, crawling,
- 36:51getting as much material as possible. And the
- 36:53sort of second and third waves that we're looking
- 36:54at now, it's more about refining and sorting
- 36:56through the immense amount of data that's out
- 36:58there. So I think we're quite good at what we
- 37:01do within the framework that we do it. As Joe
- 37:03said, that's a little bit slower than maybe the
- 37:07adrenaline rush you get from jumping on Twitter
- 37:08and firing off an ID if you think you know something.
- 37:12But what it gives people at the other end of
- 37:13that pipeline is that confident result that they
- 37:16can use in a court case or a human rights investigation
- 37:19or a really important article for the New York
- 37:20Times or CNN or whatever. So that to me is the
- 37:24focus. I think where we might expand a little
- 37:27bit in the future, and this is probably in line
- 37:28with what you're asking, is in getting a bigger
- 37:31funnel at the front. So getting more data in
- 37:32for us to process. So time and money permitting,
- 37:35that would be really cool. Yeah, exactly. I think
- 37:38the like, you know, I've been doing this a while
- 37:42now and I can most of the time I can say, okay
- 37:45Well, I think that's a hellfire right and but
- 37:47like I but I don't think that's actually of use
- 37:50to anyone in the like a there's a million, you
- 37:54know a number of variants that I might not have
- 37:56understood and like what you really want is like
- 37:59somebody like Nick or some of the people are
- 38:01from the Aries team who have like Real expertise
- 38:06and to say okay, that's definitively a hell file
- 38:08as you know, you know, there's still tentative
- 38:10but you know that like the value of the tool
- 38:13is that like the people who are verifying it
- 38:16are not You know, it's not a it's not quite like
- 38:20a community value like what we are trying to
- 38:23look into and I think we talked about this a
- 38:24lot is like trying to make the tool more community
- 38:30focused. And we've actually been really open
- 38:32to ideas about that and how we do that. Like
- 38:36how, you know, whether it be more like questions
- 38:40around, okay, well, the community, you know,
- 38:43the thing that people keep saying we're missing
- 38:45is this and some kind of process. by having that,
- 38:49I think might be more, more useful. Um, but yeah,
- 38:52like, I think the, the, like Nick said, the value
- 38:55is in having certainty that, uh, yeah, that somebody,
- 39:00you know, the methodology that like two people
- 39:03who've been trained in expert in munitions, uh,
- 39:06identification have reviewed that image rather
- 39:09than, yeah. Um, so it's a slightly different
- 39:12process, but yeah. You know, I might just take
- 39:15one more thing on the end there to say. that
- 39:17we really want this to be a useful tool. I think
- 39:20Joe said this right at the very beginning. We
- 39:21want this to be something people use. We're not
- 39:23doing this as a vanity project. No one's getting
- 39:25rich off this. So if this is something that is
- 39:28informing your work as a professional or even
- 39:30as an enthusiast, if there are ways in which
- 39:32the tool can do a better job of informing your
- 39:34work and supporting your work, we'd like to hear
- 39:36about it. That for us is probably the biggest
- 39:38community drive at the moment is how can we make
- 39:40this tool more useful? Part of that's going to
- 39:43be educating people about it, doing talks like
- 39:44this one. Part of it's going to be sitting down
- 39:47with some groups of specialized professions.
- 39:50So sitting down with some journalists, sitting
- 39:51down with some human rights researchers and saying,
- 39:53how do you use this? How could you integrate
- 39:55it into your workflow? Does it actually help?
- 39:57Like does integrating this into your workflow
- 39:58make your day slower and not help you? If so,
- 40:00we need to change something. So yeah, we're definitely
- 40:03very open to hearing from people about how we
- 40:06can improve what we're doing. The very next simple
- 40:10question comes is how do people give you that
- 40:13feedback? How do people get in touch with you?
- 40:15If they do have comments or success stories or
- 40:19things that they feel like could be improved.
- 40:24Yeah. Well, I mean, there's a, um, an email,
- 40:28which I should know off by heart, but is in the
- 40:30about page. I think it's just info at OSMP. rsmp
- 40:35at airwars .org Which yeah, and to be honest
- 40:42like We're quite open on social media. Please
- 40:45follow us on the like we've actually had some
- 40:47really good feedback on Twitter and Like blue
- 40:53sky and People and she's saying oh like even
- 40:57even like yeah, really really interesting feedback
- 41:00and saying like saying a couple of people say,
- 41:04oh, I found this image here, that look, you know,
- 41:07like when something happens, being able to use
- 41:10the OSMP as an immediate reference and then like
- 41:12linking it on social media, that kind of thing
- 41:14is really, really helpful. And so yeah, like
- 41:18I think we're pretty open to that. There's an
- 41:20email that Joe, the Joe is in charge of, but
- 41:25yeah, we'd be very open to kind of social media,
- 41:29blue sky, Twitter, et cetera. Amazing. I've just
- 41:32linked the blue sky in the chat. I did the X
- 41:35account earlier. And Izzy Bakehoven has just
- 41:40also covented the OSMP at airwas .org as the
- 41:44main place to email. So thank you so much for
- 41:46that. There's another question from Mike that
- 41:49I want to throw in. Is there a way of seeing
- 41:53known users? So for example, if someone finds
- 41:55a shell that they can... they can see the countries
- 41:58that are known to use that type of munition.
- 42:01You mentioned, for example, you can see the different
- 42:03regions in the tool. Is there a way of seeing
- 42:06different munitions that are used across regions?
- 42:10I suspect, uh, Joe Emmett might want to say something
- 42:13here as well, but I'll, I'll jump in by saying
- 42:15that's much, much, much harder than it seems
- 42:17at first blush. I think, yeah, that would be
- 42:21lovely. What I'd love to have, again, in a sort
- 42:23of perfect world is a really deep dive data page
- 42:26for each of the munition models of at least those
- 42:28of particular interest, which included, amongst
- 42:30other things, information on known users. But
- 42:33that is quite challenging to verify. So because
- 42:36we're, as Joe the Elder said, trying to keep
- 42:39to verified content and giving people confidence
- 42:42in those results, we are... probably being a
- 42:46little more stringent than maybe some other sites
- 42:49or resources in really fact -checking everything
- 42:52that goes onto those, onto those model descriptions
- 42:54and pages. So it's possible. I think it would
- 42:56be incredibly helpful, but I think it would be
- 42:58a surprisingly large and time -consuming job.
- 43:05Yeah. I'll just hop in on that. Yeah. It's, and
- 43:07Joe, sorry I passed you in a second, but yeah,
- 43:09it's something that we've talked about a lot.
- 43:12And it's not like, You can do things like, you
- 43:15know, you can search by location and specific
- 43:21munition and that filters automatically. It gets
- 43:25very complicated very quickly. I mean, yeah,
- 43:27if you take something like a Grad rocket, like
- 43:30who hasn't used a Grad rocket at this point,
- 43:33right? And so like it's easier in the context
- 43:38of Gaza, for example, where it's fairly clear.
- 43:41who is conducting airstrikes there's the app,
- 43:44but in, in other contexts, it becomes very complicated,
- 43:47very quickly. So it's something we're trying
- 43:49to work through. It's not currently available
- 43:51and it, uh, it's just something that like we
- 43:54would like to do longer term, I think. Um, yeah,
- 43:57but it's about doing it right. And, and maybe
- 44:01Joe, you've got some, uh, yeah, I just, uh, think
- 44:07re -upping what you guys have said that, um,
- 44:11with certain munitions due to the broad sway
- 44:14of actors that might be both state actors and
- 44:17non -state actors that might be using a munition,
- 44:20it won't always be necessarily as helpful as
- 44:22well in a conflict zone because the same munition
- 44:25might have been used by both sides or in the
- 44:28case of when there's multiple parties involved
- 44:31in a conflict, all of them could be using the
- 44:33munition. So it won't always be that helpful.
- 44:36If, for example, there's an Israeli airstrike
- 44:38in Syria and you have an Israeli made ammunition,
- 44:40it might be helpful in that sense. We have discussed
- 44:43it a lot, I think, as the other said, but it
- 44:46might be something for the future. But again,
- 44:47it's just, it's really hard to pin down essentially
- 44:51sometimes. Okay. We've got about 10 minutes left.
- 44:57So if you have a question, get your question
- 45:00coming in now. Quickly going back to one of the
- 45:04updates that you mentioned you were doing was
- 45:06kind of this faster response, talking about ongoing
- 45:09conflicts. And you gave a few examples, Jo, of,
- 45:13for example, the Chernobyl attack. Have you seen
- 45:17an increase of disinformation impacting your
- 45:20work a little bit with the weapons identification?
- 45:24For example, Bellingcat last year published a
- 45:27really fun article called Oh Shit. Seven deadly
- 45:31sins of open source research because we saw,
- 45:33you know, a trend of people claiming to do open
- 45:37source research and not following example rules
- 45:42that we would want people to follow. Have you
- 45:45seen a similar trend in UNITION's identification
- 45:48on social media, particularly in ongoing conflicts
- 45:51or fast moving conflicts? And how are you dealing
- 45:55with that within your own reporting on social
- 45:57media? trying to kind of get the word out there
- 46:00that you've kind of identified it successfully.
- 46:04That fills that one for Nick, I think. Yeah,
- 46:08I've got some thoughts. You know, I'm going to
- 46:11sort of, right, we're in an environment, we're
- 46:13in an information environment that's becoming
- 46:15in many ways increasingly more challenging. I
- 46:16think people know that. People are used to, you
- 46:20know, some of the disinformation campaigns that
- 46:22we've seen in recent years in particular. However,
- 46:25I think I will say that On the whole, since I've
- 46:28been poking around Twitter and online communities
- 46:31looking at arms and munitions identification,
- 46:33i .e. some time, we've got better as a whole.
- 46:37I think the level of knowledge has actually risen.
- 46:40I think people are less often wrong in a very
- 46:44public way. I know when I founded Aries, that
- 46:46was 2013. A couple of our very earliest clients
- 46:50were some NGOs and in one case a government that
- 46:53had made embarrassing public mistakes in misidentifying
- 46:56munitions. I mean embarrassing the way that actually
- 46:57affected the conclusions of their report or their
- 47:00claims about attribution in particular. So bad
- 47:02stuff. When some organizations, NGOs and other
- 47:07organizations that work in the open source space
- 47:09got started, there were some errors pretty publicly
- 47:12in some of their work as well. And a lot of that
- 47:15has been based around the fact that these groups
- 47:18often start small and with a tremendous willingness
- 47:20to assist and jump to conclusions. So I think
- 47:23as a whole, the OSINT community has matured.
- 47:25It's got better at not jumping to conclusions.
- 47:27It's got better at sort of self reflection and
- 47:31self verification and the tools that we have
- 47:34to try and combat disinformation have also improved.
- 47:36So I actually will strike an optimistic tone,
- 47:38I think, in answering that question. Amazing.
- 47:42It's actually nice to hear some optimism when
- 47:44it comes to the range of disinformation out there.
- 47:48So that's great. I wanted to also ask you a quick
- 47:51question about the 3D models, because obviously
- 47:54that's a new thing that you've introduced and
- 47:55you've spoken about it a little bit. Why was
- 47:57that decision made that the 3D models was the
- 48:00way to go forward? And what process goes into
- 48:03creating them to ensure accuracy when you're
- 48:06describing the different parts of each model?
- 48:10Yeah, sure. So we were thinking of the best way
- 48:13to essentially help visualize ammunition. And
- 48:17we thought an interactive 3D model was probably
- 48:20the best way because you can view every single
- 48:22angle. We can break it into parts and we can
- 48:25match those parts to real world examples. And
- 48:29then we could add extra detail along the side
- 48:33with it. And essentially the process is we have
- 48:41a really good a 3D modeler called Azul with Air
- 48:45Wars. And I will work with her and sort of initial
- 48:49stage. We'll discuss it with Nick and Joe about
- 48:53what is the next munitions. We have a few in
- 48:55the pipeline mainly to do with drones. We might
- 48:58be focusing on in the coming months. And what
- 49:02we aim to do is to essentially draw from technical
- 49:06documents. actually the arms manufacturers pages
- 49:11and their images of the munition, as well as
- 49:14dimensions. So on some of the future ones, we
- 49:17might potentially add dimensions to the munitions
- 49:20so you can get the correct measurements. And
- 49:23we thought, yeah, a 3D model essentially is the
- 49:25best way to go to help people understand the
- 49:28munition. Cause often we'll have images of different
- 49:31parts of the munition and maybe one image of
- 49:35Say a Gb39 hanging from or a market E4 bomb hanging
- 49:39from underneath fighter jet in one angle. But
- 49:41you wouldn't always be able to see the full thing.
- 49:44So matching these images with a model will help
- 49:47you sort of understand what the bomb looks like,
- 49:49particularly how the munition looks like, particularly
- 49:51before it's actually delivered. So before it
- 49:54impacts the target essentially. Yeah. And I would
- 49:58just add on that, that it's also like, I think
- 50:00it's really, for me, I found it, I found it really
- 50:03Yeah, useful to understand and visualize but
- 50:06also it's kind of like Taking it beyond like
- 50:10I'll just trust us to munition experts. I've
- 50:13looked at this right. It's It's like, okay You
- 50:17can you can also understand if like by linking
- 50:20to the specific images and so if you're on the
- 50:223d model You can see okay Well here is here are
- 50:26there and then it takes you back to the original
- 50:28images that we've already classified They're
- 50:30in the archive as that particular munition. So
- 50:32it's taking it beyond like Like it's an extra
- 50:36layer of kind of openness and transparency around
- 50:39almost like, you know, here is, here is this
- 50:42model. Like here are the images from our archive
- 50:45that maps these specific parts and it helps you
- 50:47both understand it. And also like, I think myself
- 50:50that like, yeah, now I, now I see things that
- 50:53I wouldn't have seen before I was working on
- 50:56this project. Um, yeah. I might, if I can quickly
- 51:00also just say that, you know, Joe said at the
- 51:02very start. that the genesis of this project
- 51:04was in many ways responding to some of those
- 51:07challenges that he had as a journalist and later
- 51:09as a researcher and indeed that I've had as a
- 51:12specialist with journalists coming to me and
- 51:14others. In fact, I can see a few people in the
- 51:16crowd here who field queries from journalists
- 51:18occasionally. So you'll probably know if you
- 51:20have an area of specialty and people ask you,
- 51:22I don't want to make this sound like we've invented
- 51:23a whole thing to make my life easier, but in
- 51:25many ways it has. We've got a tool now we can
- 51:28point people to that sort of shows them some
- 51:29of the working. You know, oftentimes, as Joe
- 51:32was saying, You know, as a journalist without
- 51:34a specialty expertise in the subject, you're
- 51:36going to struggle to get repeatability. You're
- 51:37going to struggle to get confidence in the answers.
- 51:39And you can't really check the working if you
- 51:41like, you just get a, you go to an expert and
- 51:43you say, Hey, tell me this. And they say, yep,
- 51:44it's, it's one, two, three. And you go great.
- 51:46And you take it as read. And this at least is
- 51:49just our step towards showing a bit more of the
- 51:51working out. So you'll see at the bottom of that
- 51:53GVU 39 page, for example, there's resources,
- 51:55there's links to other external sites with even
- 51:58more information and brochures and manufacturers
- 51:59information and that sort of thing. So I'd love
- 52:02us to keep going that way. Obviously they're
- 52:03quite time consuming to make, but if we can keep
- 52:05doing it, then yeah, we'd love to just sort of
- 52:08show some of the workings so people have even
- 52:09more confidence in the findings on OSMP. Amazing.
- 52:17Yeah, at Bellingcat we've also found 3D models
- 52:20really help with kind of showcasing how we've
- 52:23determined what munition is what. We found it
- 52:25really useful when comparing airstrike models.
- 52:31really quickly squeeze this question. Any plans
- 52:34for expanding the geographical coverage of the
- 52:37datasets? As someone reporting on conflict in
- 52:40Latin America and the Caribbean, this sort of
- 52:42resource would have enormous implications here.
- 52:45So in the future plans, is there any plans for
- 52:48expanding the geographical coverage of these
- 52:50datasets? Yeah, I think, sorry, I think it goes
- 52:56back to what we said earlier. Like, At the moment,
- 53:00it's restricted to explosive munitions and kind
- 53:02of heavy munitions. And Latin America tends to
- 53:04be a place where there's a lot of small arms,
- 53:07right? That may or may not change if stuff happens
- 53:12along the Mexican border, but yeah. But at the
- 53:17moment, we would cover Latin America if there
- 53:23were... like the types of munitions being used
- 53:27at scale. At the moment, it tends to be small
- 53:31arms and there is some stuff, but yeah, we definitely
- 53:34wouldn't rule out Latin America. There's also
- 53:37some other gaps. It's not that we've deliberately
- 53:41ignored certain places. It's just that we've
- 53:44focused on the places where there's the... heaviest
- 53:46use of kind of these kinds of weapons. And yeah,
- 53:49um, I was noticing the other day that we only
- 53:52have one image from Sudan, for example. Now that's
- 53:54a place where there's been quite a lot of explosive
- 53:57munitions. And so at some point I would like
- 53:59to expand that. And so, yeah, there's no, there's
- 54:01no, um, restrictions in that sense. Yeah. Um,
- 54:06and just, just on this, I know a lot of our munition
- 54:09images are currently from Ukraine and the Middle
- 54:12East and to an extent the Caucasus. But we are
- 54:15starting to add more munitions from especially
- 54:17parts of Africa. So in the pipeline, we have
- 54:20a lot of images of the MAM -L, which is a Turkish
- 54:24made air delivered bomb by the Bayraktar TB2
- 54:27drone. And that's seen a lot of use in Ethiopia,
- 54:31Libya, as well as parts of Mali and Burkina Faso.
- 54:35So we are starting to expand the geographical
- 54:40sort of range of the munitions we are documenting,
- 54:43but it does take time. And as Joe said, there
- 54:45are certain parts of the world where it's more
- 54:47small arms rather than the munitions we're currently
- 54:52documenting. But again, we'll have to look at
- 54:53that maybe in the future. Yeah. And just to hop
- 54:56back in briefly before I let Nick speak, sorry,
- 54:59is to say that, yeah, like one of the reasons
- 55:02for that has been the airwars' geographical focus
- 55:05has historically been the Middle East and Ukraine.
- 55:08And so like there are contexts that we don't
- 55:10know that much about, which is why we're hoping
- 55:12to expand these partnerships. Somewhere like
- 55:15Myanmar, for example, like we didn't, and like
- 55:18Nick can maybe talk about like the munitions
- 55:20used in Myanmar. They're very unusual and different.
- 55:24Um, but like we worked with Myanmar, uh, witness,
- 55:28uh, to build that bit of the archive because
- 55:31like, like it wouldn't be something that we would
- 55:34necessarily be focusing on as an organization.
- 55:36So like that's kind of where you would like to
- 55:40build a kind of more collaborative tool over
- 55:42time. And yeah. Um, yeah. Sorry, Nick. No, that's
- 55:46all right. You've said essentially what I was
- 55:47going to say, which is, you know, we all have
- 55:49our own backgrounds and geographical biases and
- 55:51the areas that we've worked on. And in many cases,
- 55:54we're trying to get sort of bang for buck in
- 55:56terms of the time and the funds that we have
- 55:58that are both limited for this project. So we're
- 56:01trying to, you know, look at in some ways the
- 56:03low hanging fruit and gather as much solid materials
- 56:06we can to put through the pipeline and get into
- 56:08the OSMP and into people's hands. If for no other
- 56:10reason than as Joe said before, it's important
- 56:12to scale this up and show the kind of aggregate
- 56:15value of the archive. But as we push out into
- 56:19the future, and if you're sitting there listening
- 56:21to this and you've got a couple of hundred thousand
- 56:22bucks you want to send our way, that would help.
- 56:24But as we push into the future, obviously we'd
- 56:26love to bring on people with different expertise
- 56:28and partner with organizations with different
- 56:30expertise so we can expand both the geographic
- 56:32coverage, but also the types and munitions and
- 56:34maybe the types of details that we're giving
- 56:36people as well. Love that little plug there.
- 56:39That was fantastic. Yeah, there's quite a lot
- 56:41of love in the chat. for their plan to expand,
- 56:45particularly in Africa. Subtle Knife says, Sudan,
- 56:48yes. Uri says, I'd love to see more from Mali,
- 56:52Burkina Faso, for example. And somebody else
- 56:55has said that they can help with Sudan. And they'll
- 56:58definitely email open source Neosha's portal
- 57:01now. So thanks so much for that. It's towards
- 57:05the end of the hour. Unfortunately, we do need
- 57:08to wrap. I can see people popping questions in
- 57:11still. So if the team Doesn't mind kind of staying
- 57:14a little bit just to answer questions in the
- 57:16chat. I'm sure our team would really appreciate
- 57:19that. But thank you so much for joining us today,
- 57:25Joe, Joe and Nick. It's been a pleasure to hear
- 57:28about this incredible project that you've launched
- 57:30and have been working on for so many years now.
- 57:33It's been absolutely vital to so many incredible
- 57:36investigations that have gone on over those past
- 57:40years. Yeah, thank you so much for investing
- 57:42your time there and for spending some time here.
- 57:46I just wanted to mention that quickly that we
- 57:49have in two weeks time, we're continuing this
- 57:51conversation on munitions and identification.
- 57:55And we're actually going to have a talk from
- 57:58a previous spelling cat fellow called Andrew
- 58:01Matthewson. He actually created a fantastic key
- 58:05guide that can help you identify the most common
- 58:09types of munitions that you might come across,
- 58:12particularly explosive ordinances. You can find
- 58:16that guide on the Bellingcat website, but we're
- 58:19actually going to have a chat from him in two
- 58:21weeks' time and digging in a little bit more
- 58:23on the regional differences between weaponry
- 58:27and munitions. and what kind of keys and guides
- 58:31you can create yourself or the other community
- 58:35to try and help identify these things. So please
- 58:37join us for that call. But yeah, thank you so
- 58:40much, Doe, Doe and Nick for your time today.
- 58:45It was a pleasure. Thanks, Charlie. Thank you.
- 58:48Thank you. Thank you so much all. Thank you for
- 58:51listening to the stage talk. If you'd like to
- 58:54catch a stage tour live where you can ask the
- 58:57guest questions, join the Bellingcat Discord
- 58:59server by visiting www .discord .gg slash Bellingcat.
- 59:07The music you've heard is titled Dawn by Newer
- 59:10Self and is courtesy of Artlist.