In this episode of From the Source, host Matthew Stibbe speaks with Chris P, Head of Intelligence at Blackdot, about his diverse career in military intelligence, the National Crime Agency and the challenges of big data in intelligence analysis. They discuss the importance of network analysis, the role of AI in OSINT, and the need for sovereignty in intelligence tools. Chris emphasises the limitations of OSINT and the need to use the right tools for specific questions, while also advocating for human oversight in AI applications.
Please note that any views expressed in this podcast episode are the speakers' own, and do not necessarily reflect those of Blackdot.
Matthew Stibbe (00:01.656)
Hello and welcome to From the Source, the Blackdot podcast. I'm your host Matthew Stibbe and today I'm talking to Chris P, Head of Intelligence at Blackdot itself. Great to have you on the show, Chris. Yeah, so we've been having a great conversation so far and I learned that you're new in post at Blackdot, so tell me how are you finding it?
Chris (00:13.829)
Great to be here, Matthew.
Chris (00:26.417)
Brilliant, yeah, it's hard to believe that I've only been here for three weeks. The people are lovely, the place is great. They've just made me feel really welcome. So the last three weeks have gone by very quickly, which is really positive.
Matthew Stibbe (00:40.214)
That's great to hear. Our experiences as well, lovely people and fascinating company. So tell me a little bit about your career so far. What brought you to Blackdot?
Chris (00:51.409)
Blackdot? So I've kind of had three career strands. So the first one is the military. I joined the military when I was 17, back when you could. So I joined as an army reservist back in 2005. I did five years with the Royal Anglian Regiment in the infantry and then moved over to military intelligence where I was for 16 years and still am. I've been really fortunate in that time to deploy to some really interesting places. I went to Kosovo in 2017. I went to Iraq in 2020
and Afghanistan in 2013. So I've had a really interesting military career. I'm still a serving reservist. And in fact, Blackdot are going through the Employer Recognition Scheme, part of the Armed Forces Covenant, which is really exciting. And it just shows their commitment to veterans and reservists as well. So my second career, I joined the National Crime Agency in 2014. I did six years as a...
investigator in the National Cyber Crime Unit before moving into human intelligence collection. And then after leaving the NCA, I went to a AI startup company called Logically, which specialises in mis, dis and malinformation monitoring. Really, really fascinating mission, really worthwhile mission, targeting online harm to the UK. I then went to a company called Tadaweb, where I finished as the defense and intelligence manager for the UK, responsible for
a very large account of open source intelligence users across UK government. And that was a really fascinating place, really getting to grips with how software as a service works, enabling government intelligence analysts to do their job better. And then I'm at Blackdot, which is really exciting. Yeah.
Matthew Stibbe (02:39.502)
Fascinating. And now we have to come to the really critical question, right? Is it OSINT or 'OSINT'?
Chris (02:48.096)
This is a hill I will happily die on. So I've been doing 'OSINT' since at least 2010, probably a little bit before that. And it must have been in the very first course that I sat in on that someone said 'OSINT' and it has just stuck with me. So I normally get the odd look when I'm in a meeting and people are talking about OSINT and then I'll talk about 'OSINT' and I'll get a bit of a sideways glance, but.
Yes, I will say, wasn't.
Matthew Stibbe (03:18.646)
Okay, I've always heard OSINT, but I, so we can agree to differ the reason why people could agree to differ But I'm sure in the end you will... you'll prevail. So you were telling me earlier that we've been also I know you're interested in mountaineering a very impressive picture on your LinkedIn profile We've been mispronouncing Everest as well all these years.
Chris (03:40.273)
Yeah, so that dates back a very long story, I'll shorten it, but basically it was named after a guy back in the UK called 'Everest', but when the Surveyor General named Peak 15 back in 1852, obviously the word went back via Telegram. When people started reading it in the UK, people just pronounced it as Everest and it's stuck ever since.
Matthew Stibbe (04:06.584)
So there we go, you've got two missions ahead of you, getting people to pronounce the mountain right and OSINT, 'OSINT' right. But that segues quite neatly into something that we were discussing earlier, which is actually people's names and the volume of data. And I am blessed or cursed with a...
Chris (04:08.207)
Hahaha
Matthew Stibbe (04:24.566)
a rare and unusual name. My surname is a Dutch, an unusual Dutch name. And there are, to my knowledge, three Matthew Stibbes in the world. There may be more, but one of them's a rap producer in LA and another one's a historian, and then there's little old me. But you were saying that that nowadays, if you were doing investigations, there might be thousands. So tell me a little bit more about that, this data proliferation challenge.
Chris (04:49.105)
Yeah, think, Yeah, big data is a massive issue facing intelligence analysts and investigators. Again, when I started, like you said, if we were after an individual, chances are we'd find them quite easily. Now, in the UK, there's something like 79 to 89 % of the population are on social media. And that makes it a much vaster kind of pool of people to dig through. So I think big data is such a massive issue. And that's just talking about social media content.
You know, before we even look at other data providers, for example, ad tech data or AIS data and the amount of data points now that investigators, intelligence analysts can be presented with makes it really difficult to sort through all that and come up with that, that, know, the golden nugget, the key piece of intelligence.
Matthew Stibbe (05:37.132)
And how do you tackle that?
Chris (05:41.635)
Yeah, so I'm a very visual person. I love a good network analysis chart. One of the reasons that drew me to Blackdot is how good their platform is at visually displaying that data. And I think it's really easy then for the analyst to sift through and identify key entities, key relationships, key attributes between individuals and really look at that network and kind of exploit it to its best possibility.
So yeah, I think the ability to kind of get into the data and sift it down to the, you know, like I said, that golden nugget is really important.
Matthew Stibbe (06:21.762)
How big is the haystack and how small is the golden nugget these days and what's the trend with that?
Chris (06:28.505)
Yeah, I think the haystack is now a haystack in a field of haystacks in a country full of fields. That's the problem and the country is getting bigger all the time. Looking back at statistics, I think this is a Meltwater digital trends of the UK report that said that most people have about six to seven social media profiles now. Like I said, 70 to 80 % of the population of the UK are on social media.
That makes it such a vast pool to sift through. So I think any key selector that you can get to kind of narrow down those results really helps. And I think really good intelligence tradecraft really comes into its own then. The ability to create really good structured Boolean search queries, use Google Dorks to their best ability to really hone down those results. Any context you can add will just
just help you sift through that data.
Matthew Stibbe (07:30.126)
Forgive my ignorance. Google Dorks is a phrase I haven't heard before. Tell me about that.
Chris (07:33.905)
Yeah, so Boolean search queries, and, or, all those kind of operators. Google Dorks are then kind of like another layer on top of that. So for example, you could do file type colon PDF. So within a website, it will pull back all PDF documents. I've used it before when I've done some kind of exploitation testing on websites where people have inadvertently uploaded
Excel documents that contain usernames and passwords onto the internet. You won't find that just through a Google search query, but if you use the Google Dorks, that really does narrow down those results.
Matthew Stibbe (08:12.526)
Oh, sorry, I'm familiar with that where I would search for a phrase and cite colon and then the URLs are searching with it. OK, right, fantastic. OK, good. And that volume is one issue, but we were discussing earlier where you might have
Chris (08:18.479)
Exactly that. That's the Google Talk pit.
Matthew Stibbe (08:32.136)
a proliferation of data like somebody is associated with 100 companies and you're interested in that individual and while that's not millions of data points, it's enough to expand the workload of the investigator to the point where they... Can you give me some other examples of data volume getting in the way or becoming increasingly challenging?
Chris (08:53.403)
Yeah, I was doing an investigation... So the example you gave was a Russian oligarch. I was doing an investigation last week, my second week at Blackdot and I was putting the platform to the test looking at Russian shadow fleet. So I'd identified a vessel, which was linked to sanctioned activity. It was going in and out of Russian ports. And I thought, well, let's have a look at this ship. Let's see what this comes back to.
That one ship was linked to a holding company. The holding company had roughly 50 other oil tankers. And it's then the laborious process of having a look at all of those different oil tankers, trying to work out where they've been, where they're going, what they're carrying, how old they are, see whether they fit the profile of a Russian shadow fleet vessel. And again, Blackdot's really good in terms of, because it can visualise that data.
As an intelligence analyst, can sort through it lot quicker than if it was just presented in a block of text. So yeah, that's another example of a volume of data that is difficult to sort through.
Matthew Stibbe (09:50.83)
Hmm.
Matthew Stibbe (09:56.063)
As human beings we're not very good at actually parsing lists. One's attention span and attention to detail dwindles very rapidly.
Chris (10:03.033)
Yeah, And that's why network analysis is really important. So one of the key things about any network analysis chart is that it's relatable to a human being so that you can look at a chart and say, that's a family tree or that's a company hierarchy without knowing what the context is. You can kind of visualise that straight away. And that's why network analysis is so important.
Matthew Stibbe (10:25.422)
So tell me, you've got this amazing sort of new, fresh perspective of a new job and a new year. What are you geeking out about at the moment?
Chris (10:34.959)
Yeah, a couple of things. So I think the first one is the National Police Service. As an ex-NCA officer, I'm really excited and interested to see how that goes. I think it's really interesting that there's going to be this national body responsible for national, serious organised crime, counterterrorism, bringing all those parts of law enforcement under one umbrella. I think that's really exciting. They're also going to be responsible for pushing technology forwards.
And that's something I think that law enforcement really needs, that investment in technology to be one step ahead of... Almost all criminal activity nowadays is either technology enabled or technology dependent. So I think that's something that's really quite interesting. think 2026 is going to be an interesting year geopolitically. We are only 29 days in and it's already been a fascinating year so far.
Matthew Stibbe (11:31.438)
Fascinating is one word for it, isn't it? It's A bit of a roller coaster.
Chris (11:34.289)
It is a roller coaster. I think, again, as OSINT analysts and investigators, trying to keep ahead of that is, it's really difficult because the world is changing constantly. And I think that then goes onto my third thing, is sovereignty. So I think over the past six months, sovereignty has come up more in conversations I've had with, with leaders of government departments and private industry where
you know, we need to take that inward look and think we need to be producing this ourselves. We can't rely on other nations or external bodies to provide this for us. We need to have this sovereign capability, whether that's AI sovereign capability, any other kind of technology, you know, sovereign capability. And I think that that's really interesting that we are starting to focus on the investment
within the UK to make us the leader in lot of these sectors again.
Matthew Stibbe (12:33.87)
This is obviously important on a national level and you need bodies like the new police organisation that have the kind of budgets to invest in the things that they need. But it's also actually sovereignty for businesses and managing and thinking about their dependencies. When companies are thinking about this, what do they need to start thinking about? What does our audience need to be thinking about in the world of OSINT?
Chris (12:43.558)
Mm-hmm.
Matthew Stibbe (13:03.631)
OSINT? To review and consider their dependencies and sovereignty.
Chris (13:08.527)
Yeah, I think their tooling. I think if they're using tools, and there are a plethora of OSINT tools out there that people can use, both free and paid for, maybe consider where those tools are based. Some are based abroad, some are based in the UK. Blackdot, for example, is headquartered in the UK. The developers are in the UK. It is a sovereign product, I think, by that definition.
And that's really quite important for me now. think like we said, going into 2026 is an interesting year geopolitically. That's really, really important for me. And I think having any kind of business resilience plan, if the internet was shut off, like, know, what is your business resilience plan? So yeah, I think there's quite a few things that people need to think about when it comes to sovereignty.
Matthew Stibbe (14:04.526)
A lot of risks suddenly came into a sharper focus and a lot of remote but serious risks have sort of appeared on the landscape that weren't here even a month ago or a year ago. I had the, me and my management team this week like having this discussion as a management team was like well what risks do we need to be looking at and what if for example everything on ChatGPT had
Chris (14:25.233)
Mm.
Matthew Stibbe (14:30.83)
to be disclosed or something. I'm not being paranoid. There's a danger of being a bit too tin-foil hat about it, and a little bit neurotic. But it needs thinking about. I think people need to be looking at their potential exposures and readjusting their risk assessments. Anyway, the other thing that's obviously going on and accelerating is the use and adoption of AI.
Chris (14:47.014)
Yeah.
Matthew Stibbe (14:56.846)
I'm interested in how that you think that's affecting the world of OSINT.
Chris (15:01.521)
Thank you. It's really interesting. It's, and you'll have to forgive me for the citation, so I hope it's correct, especially with my background in mis, dis and malinformation. But I read a report that said 90 % of the online content that people are exposed to by the end of this year will be AI generated. You only have to go onto TikTok, Instagram, X to see the volume of AI-generated content, either AI-edited AI-edited
AI-edited or just completely AI-generated. And in fact, there are Instagram influencers out there with hundreds of thousands of followers and they are an AI-generated individual. And that's, that's madness. That's generating revenue for that individual, which, is a lot. And I think that, you know, the problem is, is AI, AI is a fantastic tool. And I think its ability to, to hone down on the results is really important,
to consider sources that someone may not normally consider is really important. But I will always be an advocate for human in the loop. think the idea of the human having that, the accountability, the review of whatever material is collected is really important. AI can hallucinate, can make stuff up. Repeatability is an issue, I know,
within law enforcement, there's, when we talk about CPIA and the ability to repeat things, AI may give you different answers every time you ask, and that's a massive issue. There was a report from, I think it was an IBM training manual from the 1970s or 1980s that says computers aren't held accountable and therefore shouldn't be used to make any managerial decision.
How many companies out there are making managerial decisions based on AI? So it's gonna be a really interesting year. I think this will be a make or break year for AI. And we've already seen the revolution against Grok producing some fairly horrific content towards the end of last year. So it's gonna be an interesting year, I think, when it comes to AI.
Matthew Stibbe (17:10.563)
you
Matthew Stibbe (17:20.649)
Definitely.
I was reading this morning an article by Rand Fishkin who in the world of marketing, in my world, does a lot of analysis about SEO. He was working on datasets from AI tools that say you can ask it the same question in 150 different ways and get 150 different answers, whether that's the best pair of headphones to buy as a gift or the best cancer hospital in LA. Each of those answers feels very credible and
thought through and believable. I mean, part of the magic and the stress of AI is how it presents itself so credibly.
Chris (17:55.494)
Yeah.
Chris (18:03.525)
I think that's really important. And you pick up a really good point about One of my issues with AI is the lack of transparency sometimes. People may ask it, what's the best cancer hospital in LA? And it will provide an answer. But you don't know how it's reached that answer. It only takes one more step to include in the prompt and show your reasoning and show how you got to this answer. At least then the human in the loop can review
review how the AI came to that conclusion and agree or disagree or discount some of the things that it's raised. I know, for example, Blackdot, when we're looking at AI, we're gonna make sure that it's transparent so that the user will always know what the AI is doing in the background, whether that's assisting on an investigation or generating prompts for helping the analyst, that transparency I think is crucial.
Matthew Stibbe (18:58.978)
Yeah, I think this thing where it's on ChatGPT anyway, it's going little icons of where it got its data from. There's a little moment where you want to do, and this is for everyday life, just, well, let's go and click on that and see if that's actually credible information. Back in the day when I was a freelance journalist, I was writing for really very well edited, well run magazines like Wired. And they used to fact check
stuff. You'd write up, you know, 150 word little blip article for the front of the book and you get on the phone with the fact checker and they'd be asking you ... What can you give us two independent sources for that statement? You know on some German motorways there's no speed limit. I was like well I talked to the transport guy at the German Embassy. Well, what's his number? We'll give him a call, you know that level of validation, but we don't do that now I think anyway, so...
Chris (19:37.029)
Yeah.
Chris (19:45.573)
Yeah.
Yeah.
Matthew Stibbe (19:53.549)
We're coming to the close of this conversation, regrettably because I'm enjoying it very much, but I want to ask you one last question, which is if you are the consumer or the user of OSINT but not necessarily a practitioner, what sort of questions, what sort of thinking ought you to have in mind to use it wisely and well and get the best results from it?
Chris (20:16.591)
Yeah, I think there's two parts. One is about the OSINT tools at your disposal. I will always be an advocate for the right tool for the right question. If you want to display something geospatially, use a geospatial platform. If you want to investigate a network, use a network analysis platform. I'll always advocate the right tool for the right problem. I think the second one is...
is understanding the limitations of OSINT. It is not a silver bullet. I have had both my military, my civilian, national crime agency careers, people come to me and say, you know, find out this thing. I will find it if it exists. If it doesn't exist, I can't find it.
And I think that that's a really key part when it comes to OSINT. Like any other intelligence discipline, it has its limitations. You wouldn't go to an imagery intelligence analyst and ask for the image of inside a building. You know, so OSINT will provide you the answer if the answer exists. I think that's quite key.
Matthew Stibbe (21:28.92)
That's a fascinating insight. It's one of those sort of blindingly obvious, but really critical things. Back in my software development days, we always used to say, non-existent code doesn't crash. And it's the same sort of understanding. Well, thank you. That's a brilliant place, I think, to wrap this up. Chris, it's been great talking to you. Thank you very much for being on the show.
Chris (21:48.763)
Thank you.
Matthew Stibbe (21:49.551)
And that brings this episode to a close. And if you're listening and you'd like more insights into OSINT, your choice, dealer's choice, or you'd like to learn about Videris or Blackdot, please visit Blackdot Solutions and thank you for listening. Goodbye.