The Future of Open Source - A View from the CIA

Welcome to a special edition of the State Secrets Podcast. Earlier this week, say for brief CEO and publisher Susanne Kelly, welcome the director of CA's open source enterprise, Randy Nixon, to the virtual studio for a special briefing. Now, this was a fascinating conversation where Randy shared perspective on what Ocent really is, how CA uses it and its daily analytic trade craft, and a special announcement about a new tool that CA just launched to help its analysts more quickly handle the vast amounts of open source information flowing in each day. And it's typical in cyber reef events, we had a lot of really great questions and comments, including perspective from former principal deputy director of National Intelligence Sue Gordon, former deputy director of analysis for CIA Linda Weiss Gold, and chief product officer, James Ben Conklin. Here's our conversation with Randy Nixon. So just to check it out, I'd love it if you would, it's called the fight, and it is available wherever you listen to podcasts. Also, if you like those behind the scenes, type images, and pictures, I've been posting a lot of mine from my phone, right on my LinkedIn profile, from this latest trip, there was a dinner that we had that was really kind of surreal. We were meeting with a team of senior Ukrainian leaders, including the minister for strategic initiatives. We did not expect the world central kitchen humanitarian and CHAP Jose Andres to be at that dinner meeting, but he was, we'll tell you why he was there. You can follow and see some of those pictures on LinkedIn as well. And then Sean Penn, many of you may already know that he released a documentary just this month on Ukraine. He was actually there when Russia launched its full-scale invasion, talk about timing in February of 2022. So if you haven't seen the documentary, Brad and I watched it really recommended, it gives you a sense of what life is like there. So he was at that dinner as well. So if you want to see some of those pictures, be sure to check it out on LinkedIn. The other secret I want to share with you is really the best way to stay up to date on open source information that impacts US national security is by subscribing to the Cypher Briefs open source report. Brad Christian leads our incredible team on this daily report that people like Lieutenant General Bob Ashley, General John Allen, have made part of their essential daily reading. It gives you twice a day for subscriber plus members. They get open source reporting on what's happening that's impacting US national security. It used to be something that just presidents got in their daily briefings. But now you don't have to be the president to have access to this. So I really encourage you to check that out as well at the cypherbrief.com. All right, let's get some logistics out of the way for today's conversation. I'm really looking forward to this. This is an on-the-record conversation today. We have a number of Cypher brief experts and former directors of intelligence agencies on the call today. I want to thank you so much for being here with us. We also have several members of the media. So once again, it is an on-the-record briefing. For anyone who would like to ask a question, you can do that using the chat box that should be on the right hand side of your screen, if you're on a laptop or on a desktop. If you're having trouble, if you're on a mobile device, it's too difficult to figure out. Just shoot your questions via email to Brad Christian at [email protected]. You don't have to wait to ask any of your questions today, either. So please feel free to submit those at any time. Well, before we welcome our guest, I would like to thank James for their support of today's briefing. And I'd like to welcome James, Keith, product officer Ben Conclan to the Cypherbrief Studio for a few minutes. Ben, it's really nice to see you. Thanks for being here. Thanks a lot, Susanne. I really appreciate the opportunity to be here. You know, I'd really like just to get a sense from you on the private sector perspective on open source. It's impacting all of us. We're going to be talking with Randy Nixon in just a couple of minutes about how it's impacting the intelligence community. But from James' perspective, Ben, what's top of mind for you right now when you're thinking about the use and the potential use of open source intelligence? Yeah. Thanks a lot for that. I get asked this question pretty often lately, which is exciting because it's just showing you sort of the interest and appetite to learn more about open source in the community. And we, actually, matter of fact, I'm excited to do this today with Randy. Randy's one of the people I take inspiration from when I think about what we need to do to evolve our trade craft and our practice around open source. And I always like listening to his comments around the topic. I think we share a lot in common and that we both feel like people are maybe a little bit late to the party when it comes to open source. We've been doing it for a long time. I think the difference really now in sort of today's day and age is that open source has proven itself to be more than just a secondary source of intelligence. And I think as we see the conflict, the Ukraine has really told us that actually it's an ongoing persistent capability, it can bring a unique perspective to the table that maybe you don't get from other intelligence, traditional intelligence sources. And the interesting thing for me is it didn't go dark when the conflict started. I think that was something we've always sort of wondered as we think about open source. In the 225 years that James has been doing open source intelligence, we've always been focused on, you know, foundational military intelligence, specifically understanding order of battle equipment, and then really broadening that to the broader national security threat in the last few years. What we're finding is that you can really treat open source like any other collection discipline. And what I mean by that is you can focus your attention on it and actually learn things and develop intelligence just like you could say point a satellite is a specific place on the ground. That's how vast the data collection really is on open source. It's really coming down to our ability to harness that data and take it into actionable and useful intelligence information, that's really the challenge that we're facing now. So it's no longer sort of the quantity of data or the availability of data. It's not only becoming about turning that into something that we can communicate as an assured and trusted sort of piece of content and use to keep foundational intelligence up to date. For us, it's really about harnessing that. We think of it kind of three areas one, it's around structuring that information in a consistent way so you can produce results consistently for your customers. I think that's why the private sector talking, but I imagine the same thing is true in the intelligence community as well. For us, it's also about giving our analyst access to the information in an integrated way. So much like we've talked about all source and multi-ent and other disciplines. The same thing's now true in Ocent, you need to look at and understand your open source information and with satellite imagery, graphs, time, sliders, etc. So you actually need to give people an environment to experience that data. So we've had to work with our analysts to give them access to data in that way, rather than working in sort of traditional soft pipes that we might have done in the past. And then really it's around just building a consistent tradecraft that you can constantly onboard people into to continue to produce consistent results. That tradecraft is evolving constantly. I think other areas may be in traditional intelligence that we've been able to standardize tradecraft, but the nature of Ocent and the nature of the threat and how rapidly it evolves causes us to need to really keep our tradecraft fresh, keep our analysts trained and continue to expand their skills and their capabilities. All of those things I think are actually exciting. I think we see the professionalism of the Ocent community, the creation of communities around Ocent, the increased focus for media and other outlets on Ocent. We're all helping us get that higher bar of capability. We're seeing a skilled workforce and the demand for information growing. And we're seeing the trust and value of Ocent increases. So people are valuing it and pricing it and we think just the right way. So that's really kind of my summary from the James perspective. And we're really thrilled to see Randy talk today. Very important customer of ours and also hopefully really a thought leader driving the development of this for us. Well Ben, thank you so much. I really appreciate your comments. I think people should also know that you are responsible for really defining and leading the product, vision, the roadmap and the strategy at James, but you also focus on what you mentioned, interconnected intelligence, which I think is a huge aspect of this. You work with defense intelligence organizations around the world and what I think is very cool. You spent nine years as a terrain analyst in the US Marine Corps. We'll have to have a conversation about that separately. Good. Yeah, thanks a lot. Thank you. Thanks so much for being here. It's nice to have you with us. And thank you to James. Thanks. All right. It is my pleasure now to welcome Mr. Randy Nixon to the Psycho Race Studio. Mr. Nixon is the director of the CIA's open source enterprise that sits within the Directorate of Digital Innovation, also known as DDI. Now prior to this role, Mr. Nixon led DDI's digital futures team with a mission of finding private sector solutions to mission needs. So this is a fantastic conversation. You bring incredible insights both into working with a private sector and working with open source intelligence. You've also served as deputy and then director of the Office of Advanced Analytics within the Directorate of Analysis. And Randy, you've really grown up in the agency first joining in 1991 as a student intern before taking a break for a few years, you went and served in the Army where you served mostly in the Balkans, which I think we may have had some overlap there at some point. And then you rejoined the agency in 1999. Welcome. And thank you so much for being here. Well, thank you for having me. I'm honored to be here at the Cypher Brief. You know, the Cypher Brief is a really big deal in my world. And normally you have I see legends speaking here, the John McLaughlin's of the world, the Duke Gordon's, I know I'm not as entertaining as Norma Roll or Ted Singer, but I'll do my best and I really appreciate having the opportunity to join you today. Well, we're very grateful as well to have you joining us to talk about this. It's such an important issue. There is enormous interest in this briefing today because everybody wants to know what [BLANK_AUDIO] open source intelligence is and where it's headed. So why don't we just start there? Why don't you give us your definition of what open source actually is and what you think it's capable of doing for the intelligence community? - Okay, and I'll say too, if people start to see me sweat, it's my location because I bounce between buildings all day long and I'm actually sitting in my truck. It's not the questions, or maybe it's not only the questions. We'll see how we go. So open source intelligence, it starts with the simplest definition. It's publicly available information. So that's everything from the traditional media that OSC and our predecessors have been gathering since 1941 back when we were in broadcast intelligence agency where at that point, we were tasked with just finding the radio and TV, not TV, wasn't there? Our radio and newspapers out of the access powers. And it has grown over times to include, you know, television, internet, anything that's out there in the public that we can gather, social media. It's also anything that is commercially available. So if we can buy it, great literature, James, things of that nature that we are able to combine into our work force and our collection. But what's missing is our secret sauce. And the secret sauce is our data, it's our expertise of understanding our data. So our officers have a deep understanding of every source that we put into our collection. So we can give you very detailed answers about this source means this, the fact that 30% of the public of this country is reading this, it means that that, it's the technology that we've developed, which there's been a lot of press about over the last few days that delivers our collection. And then it's the insight that our officers then were able to provide the context both tactical. So helping our warfighters and our operators actually get on the X with tactical OSCENT. All the way through strategic OSCENT that supports our analysts and our policymakers. So that's a long-winded definition, but it really takes all of those pieces to make up OSCENT. And our goal for it is it's the inter-first resort. Yeah, what we mean by that is, because it is publicly commercially available, it is cheaper, easier, it's vast. And let's save our more novel ends, humans, and why not for the gaps that we can't fill our way? - And how do you look at the difference between open source that is out there domestically within the US and open source that is out there outside of the US? - Well, housed inside CA, open source enterprise, and we serve the entire intelligence community. We are in foreign intelligence organization. We are not a domestic intelligence organization. So our collection is what we've gathered from outside sources. - Got it. Do you, I know this is always a tricky question to ask anyone who works at this game, but I think it would be great if you had any kinds of examples that you could share about how the agency might use open source information and how it's affecting some of your missions without obviously getting too specific. - Yeah, I'll give you a fictional example of that's okay with you, because if I gave a real example, then that source might disappear tomorrow or move from publicly available to commercially available and we only have so much money. So in a fictional example that'll show you how we use it. So let's imagine that there is an unfriendly warlord threatening a coup in some fictional far off nation. There's our scene, so I have not said any particular country or region. So, Austin, throughout my career, and as she said, I've been doing this a very long time. So most of my career is a user of Austin. Most of my time, both in the agency army has been spent working in wars or unstable regions. I'm a junkie for that. Austin is almost always the first tip I have that there's a problem. In this example, it would be the exact same way. And often today that first hint is somebody tweeting or on telegram or sending a picture. Sometimes it's actually that press itself. But more often than not, it's that untraditional media that gives me the first clue that there's a problem. And then from that, we can immediately start alerting others in our organization and outside into the IC and to the military and start paying attention to this. Then we're going to gather all of our sources. I mentioned them before, press TV, internet, you name it. Along with companies like James would be reaching out to them and say, hey, there's a piece of equipment here that may mean we don't understand completely. What do you have on this? Of course, the available imagery. If we don't have a regular overhead of the region that's there, there's all kinds of opportunities from that point. We're going to gather together put it in some context, and we're going to give it to the director of analysis so that they can incorporate that into their executive updates and other things that they would be doing because of a breaking event as well. Because the best analysis is still that all source analysis. And we're one int that they can use as a piece of this puzzle. If this day happens to be where they only int, we're still passionate to them, working with them to produce that product is going to go to the policy maker. And this further imagine that then there's a tweet or something that alertes to who the leader of this warlord is or in people around him, we're going to dig in very technically about who the person is, where they like to be. You name it kind of juice they like, anything you can get. And that kind of information is going to be very valuable to the work fighters and the operators who would also be working on this topic. So that's why we have the full range of skills in the tactical shoot to the strategic. That kind of paints the picture about how we would piece everything together, bring our full collection against it and get it to our different customer sense. - I'm really curious to know how the agency and Sue Gordon has talked about this a lot. And I see a question from her as he, and I'm going to get to it in just one moment here. But she's talked a lot about sort of this herculean task of really sorting through all the data. And I'm wondering how you're thinking about sifting through this vast amounts of information out there because it's not like in the old days where there were maybe just a couple of sites on social media that you knew who they were. I mean, we are just inundated with information. So how are you dealing with that? - Yeah, it's our greatest weakness. And when we started again in the 40s, it was just newspapers and radio on a couple of countries that we cared about. The 50s came our own and blew our minds. And then we had television we had to worry about along with the Soviet block which had us covering the world. And you just keep marching through time and our job is getting harder and harder. It does TV to cable TV to news on those cable networks. And there was the nascent internet and there was social media big data. And you just keep going through and my officers keep getting overwhelmed. And really our greatest weakness today to our customer is the volume of what we have. And we're delivering millions of pieces of information every day, thousands and thousands of hours of video. There is absolutely no way that a human can process what we deliver. And when I say it's a weakness, it goes to the point that there are times that analysts will filter out the oceans what that we're delivering into their search engine. And then only look at the humans again and whatnot. So they're missing, they're missing those tips if they do that. That made it very important that we have the expertise and the experts that we have who can provide that context in a written product that makes it easier for the analysts and operators to understand what we're putting together and they make it sizable. But we finally reached that margin time where technology is helping us solve this problem. AI, machine translation, AI enabled other tools including summarization, sentiment, whatnot is being a complete change to the ocean community. We today no longer have that weakness. The only thing that holds me back on collection is really the AI is having the amount of money to go buy everything that's out there that's impossible. So we have to prioritize. But today and it's our main tool went live today to the intelligence community. So we're really happy about that. Our officers are able to log on. They're able to instead of having to go to 16, 17 different databases and pooled information out by searching or the analyst at CA and elsewhere using externalization trident and still having to do searches and find things. The machine is using modern graph databases, AI translation. You can go between the different languages. And it's presenting the most the events that are breaking right now in front of you. So in a summarized manner, powered by large language models with sources, so you can still go deep into it. Linda Wisegold used to always say as kind of the DA that she wanted to be able to wallow in the data. Well, if Linda's listening, Linda, you can still do that. We just made it easier for you to find what's important by putting those sources there. But then we did make it where you can interact with the data by asking questions. We've done profit engineering training. We have tools on examples on there to help people understand. And you can ask questions of the summary. And the machine will return summarized and sourced answers to your questions still. So you can go deeper and dig into there. None of this is possible. The machine is great. But it still takes a human. I mean, there's no piece of data in our tools that we haven't understood. It's only working on our culprits of information. And that's millions of pieces of information. But we understand the sources behind it. We can explain those sources to you through the tool. And then before you trust the machine to give you the actual analysis, and you certainly don't use it to put the work out or the operating on the X, use it as a great starting point. It's that human machine team that is our present and is going to be our future as we go forward. Yeah, I think that human machine team is such a critical part of everything we're talking about right now, open source AI, you name it. But I just want to back up for one second. The tool that you were just describing is a new tool that the agency is using that's sort of like your own version of a TAC GPT with trusted sources fed into it. Am I getting the big picture right? It's not just CIA. So OSCE is responsible for-- So CIA is functional manager for two ends. People forget about mine more often than not, because human gets all the Hollywood attention. You can watch any TV show and it's about human. Director Burns also is the functional manager for Austin. and by being what we mean by essentially managers a week. are responsible for the intelligence community, not just ourselves. So this tool, the trade craft we use, and everything behind it is available to all 18 agencies of the intelligence community. And yes, chat is one small feature. But if I turned off the chat feature today, it's still mind blowing only changes. Oh, so for us, with the summarization, with the way we can use our sources and the way you can find things, it can send alerts, it'll send information to your phone, to your high side, or really, we have built this to be an easy push discovery tool. But we also still make it where, say, you're a Russian economist, you have a four-use section where it will deliver just what you're interested in, just like the analysts are used to using an excernation trident with their profiles. And you can also do searches, so you don't have to, if it's not pushed into what you want, you can search you can go in there. It's a full discovery tool, push and pull in a way that we've never been able to do before. It sounds amazing. And I kind of feel like everybody needs to have this. So I'm going to get to see Gordon's question now. What she says, we are hearing a lot about this new tool. Let's see, it is, have you thought about making that openly available? She asks, wouldn't your best, like, wouldn't you best serve the general public in this time of information disorder, which we know we all have? And she says, thanks for being here. Sue, I totally canceled the question. I'm sorry, but I think you got the gist of it, Randy. I understand what she means. Everybody wishes that we had the old OSC.gov that was much wider available to the US government and had leaked out to the general public in many ways it shouldn't have, and to academia, and a bunch of other places. Because of the easy way to sign up to, you were supposed to have a.gov account, but there are so many people that have a.gov account that it was a very insecure system. So it made it a cybersecurity threat to everything. So we, right now this one is more locked down. You have to be an IC member. You have to sign up on the high side systems to get access to this low-side system. You have to have two-factor authentication through not your phone, because we can't take our phone into work, or I wouldn't be doing this from my truck, but two-factor authentication to get the access to it. Ideally, yes, everything she says is right. My first challenge is to get a two non-IC members. We have other tools where we deliver information to out the wider US government today that aren't nearly as good, that we want to upgrade and make more compatible with this. I would love to give access wider to users on our exact -- this actual tool, but we need to have a better identity management system ready to go, and we're working on that. But I am not foolish enough to give anybody a timeframe when I think it will be delivered. So the intention is to expand our users, or at minimum, to build parallel systems that can talk to each other that have these same kind of capabilities. But, you know, this is phase one. Phase one day one, so we'll get there. Everybody wants everything in phase one. You know this. Why don't we? There's a feedback session in section of our tool, and I think the number one user of saying, "Hey, I need this. It's me, it's my folks." That's good. Okay, so you mentioned Linda Weisgold, who we're also very proud as a Cypher brief expert, and Linda has a question for you. She says, "Would you please ask Randy about your view on creating a new separate open source agency?" She says, "I'm not in favor of doing so because I believe analysts need to wallow in the data as you mentioned. He's becoming known for this phrase now." And a separate agency really risks putting obstacles between analysis and data in this video. What do you think? Yeah, I think there's been -- I love that people talk about it because it keeps us in the press. And you keep us in the press. The public, Congress, others understanding Ocent better is amazing. It's very helpful to me. So, I'm grateful to the AMUZG rights in the world who are writing books and articles on it. And if somebody I talk to a lot as well. But I think saying any way agencies an easy button, because people think human has one agency has got CIA. Other places do some human too, but CIA functional management runs it. And there's -- say, again, there's NSA, NGA, and Fisher Cardillo, I believe is here. You know, NGA, he knows -- they have the geo in it. So why isn't there one? Ocent, it's an easy way to talk about it. Most of those people completely -- and not in a bad way, but they're ignorant of how we do operate across the US government in this federated model. And I'll explain a little bit about that so you can see how we actually are needed together. And then what I think is the strength of that. So as I mentioned earlier, Dr. Burns is functional manager of this. He runs it through an organization called the Board of Governors. It's Dr. Burns chaired. It's -- it's people of his stature across the I.C. who serve on this body to oversee Ocent across the community. He delegates the -- and we meet twice a year as a Board of Governors. He delegates that day-to-day authority down to open source enterprise. So to me, I take myself as the living embodiment of Ocent. That's why I'm the salesman for it as we go forward and help drive it. We run in across the community on something called the National Open Source Committee. The NASC is what we show in that for. We have been meeting monthly throughout the -- since I joined OSE in December. By charter, we have to meet every other month. But there's so much work to do. We've been meeting with a greater battle rhythm. And under that, we have subcommittees where we actually have the technical experts and the collection experts on why not working on everything from trade craft to data and collection to make sure we buy once, share with all. And so forth working these big issues. Across that, that knits together the community. So it's OSE sitting right next to my DIA colleague who actually we talk almost every day with NSA and all the other agencies so we're working these problems together. So it's not just my resources. It's our resources. Now we move for part together as one. And Linda's right. If we did open up tomorrow and Congress created a new agency and we have an OSE community, I think it's a fallacy that we would save resources because C.I. is not going to take the OSE and away. And from being right there in the mission center sitting with our analysts and operators, D.I., certainly wouldn't do that as well. We would be doing duplication of effort. And we are better when we're sitting and co-located with our primary mission partners to accomplish what we need to do for the US government. Now, and I'll always end with the joke, you know, I woke up tomorrow and they did create a new agency. Go ahead and put me in charge of it. I'll be alright that. That would be okay. I get this really interesting breakdown. And the most detailed breakdown I've heard of why you feel the way you do about that and why others. So like Linda probably feel the same way. So I really appreciate that. We've gotten quite a few questions related to potential ethical considerations of using publicly available information, such as some international private sector or cell phone type data. I think the issue really here relates to potential privacy concerns for Americans. How are you thinking about that? It's paramount to everything we do. My corporate board has a lawyer who sits on it, who's involved in everything we do. Congress is very involved and they ask this exact same question a lot. I've been on the hill a lot throughout my career, usually talking about some more conflict. I've never gone down with more lawyers than when we talk about privacy and US public's privacy in particular. And as an American, I'm very happy that that's the case because I want my privacy protected as well. People need to understand that, you know, we do have very detailed laws. We have to get old guidance and come from D&I and our own organization. Our lawyers keep us very honest on that. And if any, if, you know, Randy Spidey sends ever goes off on some weird thing we're trying to do to get collection, the first thing I do is bring in the lawyer and my counterintelligence officer. So we all walk through it all together. So I can't go into minute details because then it stops being unclassified. But I love to tell you that how seriously we take it, how seriously our overseers at Hipsy and Sissy take it and how we interact with them and the legal community to make sure that we are staying on the right side of the law and guidance at all times. And that is fair amount to everything we do. >> Okay. We have a question here from Morgan Bryant, who asked, can you talk a little bit more about the role of AI and other machine learning models in Ocent Collection and Analysis? >> Yeah. The very, we've been doing AI since the very beginning. OSE is one of the pioneers in the space for two reasons. One, we're operating on the on-class side of systems, so it's a little bit easier than when we're up on J-Wayx or the high side. So we can often be the test base before we move things to other computer systems. So, and the other reason is machine translation, AI and the ability for machines to do translation will be one of the first real use cases for this technology. And because of what we do, it became very important to us very quickly. The important thing we've done is, you know, there's my folks speak over 60 languages themselves. So we have a lot of expertise in language stick abilities as it is. It helps us do that human machine partnership I was talking about, so we can do it in a large scale, but we can test it. It also was one of the volume problems, because when we can translate everything by machine and at least have it be good enough that we could share it massively through the scale of what we were delivering as well. So it's made the people part that much more important. And now the AI and the summarization efforts they can do with the larger, much models, that much more important, as we are able to cobble together information in a way that helps us return time to deeper thinking, which is really what we've done. What I mean by that is, you know, besides having to translate everything by hand, we don't have to do that anymore. We have to check it. And our new machine helps you to toggle between the two languages easier, makes it easier to check the sources as well. But we also used to have to hand, right? And use brain power to produce summaries ourselves or in ashore. action pieces. And when we started to roll out this new technology, we put together a working group of officers, not just an OSE, but across the community to look at the benefits, the costs, and how these machines work, and how we can use it. And universally, the practitioners were the main people on this, so my open source specialist. And the day themselves came up with it, once we've fully delivered the tool to everybody, we don't have to do summaries anymore. That saves us hours a day to be doing them more deep thinking, because the machine is doing it, and it's putting it right there for it in a way that the users can check the sources and everything themselves. So we get the return time to deeper context and deeper, and so that Linda doesn't have a problem. Deeper single source analysis. We don't do multiple source analysis in OSE. Cynthia Strand, I think you know her. She asks, how do you anticipate giving all source analysts the ability to explore data in the wild of the internet directly to assess that raw information against classified sources? We've been doing that for years. Now first, this new tool that I've explained, we are going to make available to the director of analysis widely. We've been working towards making being a lifetime analyst for the most of my career, anyhow, I understood very early on that OSE was holding information, that we didn't have an easy way to deliver. Next generation tried in our main search engine is great for textile products. The mapping tool that OSE or the DA uses, that OA runs, the Office ofance Analytics, I helped develop that tool. It's great for some of the other things that's not textile, but the best scale of video we produce, there was no good way for the DA to get that. A lot of our pictures, no great way for the DA to get that. So one of the first things we've been doing over the last year and a half is going ahead and giving DA analyst access to what had previously been open source tools so that they could easily access this type of information. While on the data, that much easier. We've just given a whole bunch more data at that point. The new machine and the new practice just makes it that much easier for them. Now if I go to a video and I'm wanting to know what's going on in Ukraine and I want to see what's the newest thing on Mr. Slinsky or President Slinsky, it'll take you right to that section of the video. You don't have to even scan through it and it'll automatically trains like that video as well. So you have the text right there beside you too. So we are opening up our collection in a way that was never possible because of what technology is making possible today. Repositing our conversation for this message. A lot of experts who read the Cypher brief are former government officials who were brief daily on national security related news around the world while they were in government. But when they retired and lost access to those briefings, they needed a new source for late breaking news that impacts national security. So the Cypher brief created the Monday through Friday open source report. In just about 10 minutes a day, you'll be up to speed on the story's impacting national security from around the world. You can get your access at the cypherbrief.com. And it's moving quickly Samadiki says nowadays we're seeing a lot of misinformation and fake content in Ocent and it's getting more difficult to discern what's real and what's fake. Plus he says add to that high quality impersonation. How much effort is being put into validating reality versus fake content at scale. And we do have programs on that. I can't go deeply into them. It is certainly a challenge and it's something the private sector works on and they work on it often in a very interesting way. It's the companies who are producing the pictures in the video who will then test for fakes. So they're making fakes to do that. So they just made it better at the fakes. So they make their tests better when just keep going on and on and the industry is making it harder and harder and harder. This is a place where when we're providing things at scale that that mass volume is helpful because often those fakes when you're delivering millions of documents end up on the far end of the scale and we're able to weed those out quicker so we don't even put those on front of the officers. Now when they when we do discover them, we are working with the methods and other technology to be able to say, you know, I met a dad or what not, you know, determine, you know, that picture is not real or this picture also was used six months ago somewhere else. So we know that story's not there. That's where the expertise of the officers involved and the bulk of our workforce is Mission Center aligned and mission aligned. That's where their expertise comes into play to help the analysts and operators understand what they're seeing. A lot of those private sector companies said to you that kind of work or putting it out in the wild too and kind of debunking these myths as they get out there. We've had a lot of questions on this topic. How does the new capability account for analytic bias and how does the process fit within the analysis workflow? Well, Linda could give you a better answer than that to me probably. So I will go into the allied bias part. What I know our mission can do is we have a built-in where it's explained in the bias by right there in front of the officers of the product they're reading. So let's help them understand any of the biases that they may be seeing. An example of that is if I'm looking at an event and I've got a hundred pieces of press that have come in on it or video you name it. Remember pieces of Ocent. We're able to go down to the point that we can say I want to see what the Russian public itself is seeing and I can do that. We can weed it down there so that we can understand how that was being presented just to their own citizens. And then we know the other way. Show me what the Russian press is saying to the outside world, easy. And that was something that was impossible without hours and hours and hours of work in the past. So and then there's a sliding scale. So officers can see that a 40% of your sources that you're looking at at this previous picture right now are biased and they have a slam from this government or whatnot. So then you can take those sources out and get a less biased answer. So then that helps the analyst present a view that is all encompassing or how are they needed to explain it. It makes it easier for our Ocent to be used. There was a second part of that question that I totally forgot. So how does the process fit? Yeah, the workflow. Yeah, we're right now. We're working on that. Now most of the tools we build, there's a very there's an easy button. If you're looking at a screen, you had a source it button and it loads it up into the what the analysts are writing on. And with the sources just ought to populate. Because our tool is on the low side, we haven't hooked it up to those systems yet. Today is going to stay one. But it is on our work plan. So the and the first thing we had to do in large, if you use cat GPT or bar or any of the language language models at home, it can do great things. It can write poems for you. You can do scripts songs. I made it write a song about Ocent once. It wasn't that good, but it was still a song about Ocent. Because Hollywood's never going to do that for me. But but it doesn't source. So you don't know what you're looking at. So the first and the most important thing to make it usable for the director of analysis and even for ourselves, machine had to tell us how did you make this summary and what did it come from. And then so I can go check it myself. And we made that possible very quickly. And let's say quickly this program, this build started six months ago. Wow. The technology been very fast. And the private sector's partnership, it makes one of this possible. And I know we're going to talk more about how we work with private sector a little bit. I do want to talk about private sector because I do feel like that's a huge issue. I have just a couple more questions I want to get to first. And then I'm going to turn the focus of the conversation on that. One one question here, Mr. Nixon, as an Ocent junkie, have you come across analysis or policymakers who tend to downplay Ocent or see it as a source of less importance than classified and promotion? And if so, how have you enlightened them on the benefits of Ocent and advocated for this capability? Because you've said yourself, you're the best salesman for this. Yeah. I think once the and the new administrations come in, often the more sexy ants, and I bet you didn't expect that word to be used today, but the more sexy ants or what draw their attention because it's novel to them. Humans awesome. But once they become more mature and that doesn't take very long because they're reading their junkies themselves. So they wouldn't be in those positions. They understand the power of all source and all source and the importance of Ocent in that. And the very first time they have a crisis and I'm and Ocent, the only thing they have, it helps them grab onto it very quickly. But the first question I get from almost everybody is, how can I tap more, which goes back to Sue's question about how do we deliver more of this to a wider scope of people? There's a hunger for it. The hunger comes a lot of it from the things like this that raise the attention of what Ocent is. And we can point them to some good sites that they can go to and for some people in the wider Ocent community, we make sure that they get what they need so they can deliver it to their individual policy makes as well. Yeah. Okay. So open source is an inch as we keep talking about these intelligence sort of inputs that the intelligence community uses. Human intelligence sources, SIGINT signals, an open source is Ocent. So it's a zone four of intelligence. But how will this particular int work with private sector to ensure that it's the best product available to government decision makers? Because as you just said yourself, the pace of technology is changing so quickly even within the private sector. So how are you thinking about those partnerships and how important are they? Oh, they're they're vital and not just for Ocent, they're vital for the intelligence community writ large. And as you mentioned before, much up before this was with our organization, which was the interface between the director of digital innovation and the private sector. That meets the needs not just of of DDI, but for the entire community. And there's also places like the technology and T2MC, our missions center that faces works on this issue. We have our own CTO who's often out there meeting with companies as well. But we have lowered the barrier so that it's not just the same four or five companies that everybody thinks of that you you always work with. What we did is. We created a broad agency announcement, it's called digital hammer, and we opened that up very easily to companies across the United States at last count. It was over 300 companies across the US, not just California, there's great innovation happening in Boston and Dallas and Washington, you name it. Great ideas can be found anywhere and we're making it easier for those companies to find it. For people that are interested, if you send an email to [email protected], digital futures, that's the organization, we'll make sure that you get information about how you two can register and be in part of digital hammer. And we put out unclassified challenges out there and we put some out recently from OSCE looking for better language abilities on certain languages. We want some of the Asian languages or heritage translate, we'll put out there and say can you help us with this, or can you help us with the obvious recognition or takes and videos. And we'll put it out in an unclassified way, we give companies several weeks to a month to send us in a five page white paper, we're averaging between 15 and 25 papers every time we put out a challenge from companies across the world big to small. And we work with venture capitalists when I see small, and he's going to be very new companies all the way to ones that have been working in defense and intelligence industry since the 40s. Then we whittle down those papers, we pick the top five or three or sometimes zero depending on the answers and we ask for full proposals, we get those full proposals back in the month. From there we can prototype or not, and at that point we can ask for social justification because we've little things down properly and start to move something to our systems. So what used to take 12, 18, 24 months longer to find the new technology and about if it takes that long the tech has moved already, we can do in months now. And we're doing it broadly across, we also see myself Jennifer you bank is ahead of the DI and others were out on LinkedIn and we post these things and our challenges as well, so we're literally opening away agency officers have never been before. And you're on LinkedIn at Randall next and I think right there. I was trying to find you earlier, all right, if you were able to press a button to create your own open source intelligence product that will exist 10 years in the future where you don't. Have to worry about today's hurdles or challenges, what would that product look like? It was still involved that human machine team, but it would be even faster at this point and would be to deliver it quicker to a wider range of customers. To your cell phone or whatever exists 10 years from now, I was just reading a story about how the phones may change so to your eyepiece that you're bringing in plants, we link it to information quicker, easier with videos videos or are still hard, it's still time intensive to put those into our products. So video imagery storytelling in a way that's more like the outside and a quicker pace. The most important thing for me 10 years from now, though, is much more micro, you know, we've mentioned all the the icons of the intelligence community that you're watching today. I mentioned Jennifer, you banged, she's the only one who's still with us inside the community, there are others are alive. But we're just caused as we leave a new generation comes and the impact and the mission remains so the most important thing to me is that I've raised up the next generation of the ocean professionals and somebody like Linda, I am from Linda's coaching tree. And from Jennifer, you banged coaching tree now that there's a coaching tree that's raining Nixon that 10 years from now is running the ocean community and continuing to just strive to put tech and context together in a way to deliver mission impact. So it's if that's how we last that's our legacy. So there's a good question here that I'm going to wait for just a moment about training that is a good follow on for that. Before we get to that, you know, we haven't really even mentioned the word China in the whole time we've been talking and there's such a race for technology. There's such a competition for data how that data is used China doesn't have the same laws that the CIA has or the intelligence community has that govern the use of that data. How do you see this new tool and sort of the mindset of the intelligence community when you're looking at a pacing threat like China. We're certainly in a competition and I think that competition is a great word for it. I'm a very competitive person and second place is first loser in my world. So moving fast on technology with proper tradecraft and human expertise behind it is important because I know my counterpart sitting in Beijing is trying to do the exact same thing and information dominance gives us time and often seconds make a very big difference. So I was meeting with a large group of liaison last week and so I was talking to them about is that we can work together today sharing information again. I want to share widely of them again sharing expertise sharing regional knowledge and peace. Or we'll be back here doing it or and that's when you're sitting and we were in Seoul. I was creating and start shooting you right from where you're sitting and then trying to in traits or just down the road and meeting the country that are with or are already being impacted by the worst of your current conflict. It should resonate with everybody we have an opportunity to move fast, cooperatively across the IC and with our partners and private sector and liaison. Now so that maybe that worst case doesn't come. So bring that and one that you know while this is a fun conversation what's at risk here is less fun and very serious. I'm glad you kind of put that into perspective for us. I want to get back to that question that Brendan Moore had about training. How do you train up your top to your open source analyst for a constantly changing landscape and competitive one given what we know about China. It's difficult because it's changing who fast and places like the director analysis have a great training program we take part in pieces of that. One of the first things we did when I came in to open source enterprises create a new group is our workforce engagement group. It's job is to work with our own DDI university with DNI's university and the other school houses across the IC because of DIA or the or the army to teach and a great ocean course we just need to expand that out as a role as functional manager. We have been cataloging what's available not just from CIA but from everybody so we can build not just our experts as the question asked but we need to have a lifelong training for our new hires all the way through those experts. So that they can learn the deep trade craft we're not just going out and doing Google search just there is a deep methodological trade craft of what we do. Both on the tactical and strategic ends on how we find our sources and understand them. So we do have programs for that and we're looking to make it even better and closer and the army just lost the whole new program and we're going out to monitor that one here in November. So we can see what we can expand and build from there because again there is limited resources in all the government there's a lot of resources in osent and we want to be able to just like we want to buy. Our commercial label information wants to share whenever we can we also want to take advantage of each other's training and trade craft across the community spread the best practices and that's part of the job that director burns is passed down to us. So we take it very seriously and have a unit focus just on that. And focusing for just a minute on that unclassified component to this Matt Smith says regarding unclassified data processing models are changing quickly are you partnered with or experimenting with other models such as public cloud providers. So you know the government has its five cloud providers and we have one of the first big contracts inside of osc that is able to touch and move between those five with ease and more importantly we build our tools very modularly so if I'm doing machine translation on Hindu it may not be. So Google's who's got the best one but I can easily plug in any of those like I'm from the five clouds any that we've developed ourselves because we do if we're having to do hand translation we're turning that back into training models. So modulator and even today if the chat we're using tomorrow we wake up and it's not the best we flip it out so having our builds being a way that we can plug and play is very important to stay in on the very cost of technology. So we're really expecting like we would have with machine translation is something that we developed or a company developed or just open source and very might be the best for this language but not for that one and any be able to use all of them I think in the future. The large language model that's good for this may not be as good for that and I'm going to have both of them and there'll be a third or fifth six and we build our stuff and so that we can take advantage of that. I would like to ask about challenges and opportunities because I think it really helps people understand the big picture so what are the biggest challenges that you're facing right now and in your push to keep focus on on what open source intelligence really brings to the game when it comes to the IC. I think we've touched on most of those and we are first challenge when we started this journey together as with me and the OC LSE colleagues in December was that volume and how we couldn't get through it and I certainly had a deep understanding of that coming from the analytic side. I've been on the operation side in my past as well so understanding that didn't matter how much we collected if it was unusable and I think we've worked really hard to mitigate that problem today with what we're doing. The other is the misunderstanding from from the community and from the general public about how we do operating this federated model that we talked about earlier. and we've worked really hard to raise the intelligent to acumen of everyone about how we do cooperate and work and work very closely together. And there are a lot of eyes on us to make sure we do that from Congress and our individual ages and whatnot. And I can tell you, they're very happy with what we're moving as a community. And then we talked about training, just keeping pace and continuing to upskill our workforce. - So Brad has a follow on question to that. And he wants to know if the workforce at CIA and the broader USIC are embracing the use of open source at the pace that you need. - Oh yeah, and I can tell you that's never, it's a misnomer that Ossend isn't used. When I came back out of the arm, I used it when I was in the army. So on the front lines, sitting in Bosnia, I was using Ossend along with everything else. As a military analyst following Kosovo, I was using Ossend, you just keep moving through my career and you name the war when I was working with Ossend was paramount to what we were doing. James being one of the piece of that Ossend and no understanding the equipment needs are there doing the order of battle, which I'm still a junkie for. Ossend is and always has been and always will be incredibly available. And for me, it's making it easier for it to get in today, hands and a way that's usable to our customers and our mission partners. - Okay, last question from our attendees today. You mentioned cross-source queries as a solution to help with the challenge of the volume of data. How does this work in an environment with such a variety of data types that are in the open source space, like images, videos, temporary data, like stories, things like that? - You have to have a very modern underpending architecture to your system. Graph database is automated tagging of your information. So it's able so that AI is able to help and work on the summarization of the AI, does help us do those kind of things as well. And I'm very proud of what OSE had done well before I got there and ensuring that we had that modern underpending of our data. 'Cause the data can be very difficult to use because it has looked so different. There was a tool that we brought in for the director of analysis trying to do summarization and the first time we used it because the data sources looked so different. It told me the top word is top secret, top locations of Washington. That didn't help me. And it took quite some time to rewrite those algorithms to get it to where it was useful. OSE figured that out well before anybody else and built our architecture in a way that we're able to do it at scale. And if we couldn't, this tool wouldn't be here today because that is absolutely necessary. Data cleaning, data tagging, and having those inner continuity sort of graph databases, you know, Ben Cockleman tell you the same thing with what they're doing with Antara that underpins everything we can do. - This has been a fascinating conversation. I want everyone to know that the email address that Randy had mentioned about working with private sector. If you want to send in information is in the chat box. But I want to come to you now and just thank you so much for all of your time today. I think you've really helped us understand better what some of the challenges are, how this is working, what the opportunities are, what's at stake, why open source information is so, intelligence is so important right now. Any closing thoughts for us today? - Yep, no. I really appreciate the questions. I like the focus on OSE's is out there because it does raise the acumen, which does help all of my intelligence community. So I'm once again honored to be here. It's at the legends, I'm not worthy. The legends of AIC are what normally are on the cipher brief. I am a reader of your daily OSE material. - I wonder who that is. - So, but thank you for having me and I hope we can do some more things together in the future. - Well, thank you so much. I really appreciate it as well. Randy Nixon is the chief of the open source enterprise at CIA and it's been a fascinating conversation. Thank you for that. - Thanks. I'd also like to thank everyone else. We had hundreds and hundreds and hundreds and hundreds of people register for this event, which just gives you a sense of the appetite for open source information right now. I really want to thank James for their sponsorship of this conversation allowing this to happen. We'll see you back here again soon. In the meantime, make sure you're getting your own daily open source report at thecyfordbrief.com. And I will see you next time. - Thanks for joining us for another episode of the Cipher Brief State Secrets podcast. Until next time.