Latest / Global Insights / Spies, Satellites & Startups: Intel in the Digital Age

Transcript

Spies, Satellites & Startups: Intel in the Digital Age · Global Insights

• 0:00
  SPEAKER_03: The longer a user is on TikTok, the more benevolent view of the Chinese Communist Party's human rights record they will have.
• 0:11
  SPEAKER_01: In March, a French sailor went for a run on the deck of the Charles de Gaulle, France's only aircraft carrier.
• 0:18
  SPEAKER_01: He posted the run to a fitness app, and a few days later, Le Monde, one of France's biggest newspapers, matched it against satellite pictures and reported the carrier was off the coast of Cyprus.
• 0:30
  SPEAKER_01: The French Navy keeps that position secret.
• 0:32
  SPEAKER_01: His phone gave the carrier away.
• 0:35
  SPEAKER_01: Your phone keeps the same kind of record about you.
• 0:39
  SPEAKER_02: Your car collects everywhere you go, that data is commercially available.
• 0:44
  SPEAKER_03: Data like that can be used to target Americans.
• 0:53
  SPEAKER_01: It can also make a photo or video of something that never happened.
• 0:57
  SPEAKER_03: And I think as there's more AI in the world, you actually need more human intelligence because we, as an all people, are going to stop trusting anything we see online.
• 1:11
  SPEAKER_03: Because anything can be faked.
• 1:15
  SPEAKER_01: In July, we sat down with two people who spent their careers in American intelligence.
• 1:20
  SPEAKER_01: We asked them how spying works now that everyone can do it, and whether America is keeping up with that.
• 1:26
  SPEAKER_01: This is Global Insights by Network 2020.
• 1:30
  SPEAKER_01: Today, spies, satellites, and startups.
• 1:41
  SPEAKER_01: Joining us today is Dr.
• 1:43
  SPEAKER_01: Elliot Jardines, the first official to oversee open source intelligence across America's spy agencies, and Dr.
• 1:49
  SPEAKER_01: Anthony Vinci, the author of the Fourth Intelligence Revolution.
• 1:53
  SPEAKER_01: Our conversation is moderated by Courtney Doggard, the president of Network 2020.
• 2:01
  SPEAKER_00: Almost everyone now has this exquisite tracking and data device that is also a font of information in an era where we now have the technology to sift through all of this.
• 2:11
  SPEAKER_00: In the Cold War, a spy might recruit an official, steal a file, tap a communications line.
• 2:17
  SPEAKER_00: Today, countries pursue the same objectives of access, influence, and disruption and advantage, but today their operations are more widespread as global data networks, commercial satellites, cyber systems, social platforms, and AI are all operating domains.
• 2:35
  SPEAKER_00: A hostile service can reach further, move faster, and hide inside environments that no government fully controls.
• 2:42
  SPEAKER_00: So, how can the US and allies defend themselves?
• 2:45
  SPEAKER_00: How do companies, as both targets and sources of information, work with governments?
• 2:50
  SPEAKER_00: And in an era where speed is paramount, how do we construct appropriate guardrails that maintain both trust and effectiveness?
• 2:58
  SPEAKER_00: Anthony, I'd love to start with you because you've written a book on the subject.
• 3:02
  SPEAKER_00: In the fourth intelligence revolution, you trace intelligence from World War II espionage through the spy versus spy competition of the Cold War and then counterterrorism missions after 9-11 to the world we're entering in now.
• 3:18
  SPEAKER_00: So, what has changed so fundamentally that you call this a fourth revolution?
• 3:23
  SPEAKER_00: And what does espionage look like going forward?
• 3:28
  SPEAKER_03: It's helpful when thinking about intelligence to kind of look at it historically.
• 3:32
  SPEAKER_03: I think that there's this tendency for many of us to sort of have an image of intelligence.
• 3:37
  SPEAKER_03: And a lot of that, frankly, is based on the Cold War.
• 3:40
  SPEAKER_03: We sort of imagine James Bond, we think about the CIA and the KGB.
• 3:44
  SPEAKER_03: We, you know, picture a case officer.
• 3:46
  SPEAKER_03: I used to be a case officer at a diplomatic reception at an embassy or something, trying to recruit some, you know, under-secretary of something for in a foreign country.
• 3:57
  SPEAKER_03: That world still exists, but that was a snapshot in time.
• 4:00
  SPEAKER_03: In America, at least, where we started with intelligence earlier in our history, is intelligence was primarily a military issue.
• 4:08
  SPEAKER_03: Um we didn't really have a civilian nor a peacetime intelligence agency for most of our history.
• 4:14
  SPEAKER_03: Um we had military intelligence run by the different services, um, which did not report to the president.
• 4:20
  SPEAKER_03: And then World War II, we started to professionalize intelligence more.
• 4:24
  SPEAKER_03: We created the OSS, which centralized intelligence to the president and made it more of what we imagine today.
• 4:31
  SPEAKER_03: The Cold War changed intelligence again and really professionalized it and created the CIA and DIA and NSA and other agencies, increased and professionalized the intelligence workforce and really focused it on political and military issues.
• 4:48
  SPEAKER_03: But it was also made it very, very secret.
• 4:51
  SPEAKER_03: That's what many of us imagine as this highly compartmented, very, very secret activity where most Americans were not involved in intelligence whatsoever, and only occasionally read about it in a news story.
• 5:06
  SPEAKER_03: Um, and usually when something went wrong, not when it weren't right.
• 5:10
  SPEAKER_03: 9-11 started to change that.
• 5:12
  SPEAKER_03: And all of a sudden, we needed other people involved in intelligence.
• 5:18
  SPEAKER_03: We um had other government agencies involved.
• 5:20
  SPEAKER_03: So um it was called a whole of government approach.
• 5:23
  SPEAKER_03: We created the Office of the Director of National Intelligence to kind of oversee an intelligence community, we called it.
• 5:30
  SPEAKER_03: And so all of a sudden, FBI agents and even people working at the Department of Transportation were kind of incorporated into the intelligence world.
• 5:40
  SPEAKER_03: And intelligence agencies were starting to use more commercial technology.
• 5:46
  SPEAKER_03: In fact, the CIA had created a venture capital organization called InkyTel in in the late 90s and really grew that after 9-11 to bring in um Silicon Valley type technology.
• 5:58
  SPEAKER_03: So all of a sudden, you're working at a technology startup and you might actually end up working with the CIA in some way, um, which was very new and very different from a world where we had Lockheed Martin and companies like that um building secret spy planes.
• 6:12
  SPEAKER_03: And so um we also started to do open source intelligence.
• 6:16
  SPEAKER_03: And and Elliot um really is the expert here to talk about that.
• 6:20
  SPEAKER_03: But that was where the government not only realized we needed this technology from um um these other agencies, but also that we could actually perform in intelligence publicly and look at public sources and bring that in.
• 6:35
  SPEAKER_03: OSS had done this in World War II.
• 6:38
  SPEAKER_03: Intelligence agencies had always read local newspapers in the Soviet Union, for example, like we had always done this, but now we're doing it at a much larger scale.
• 6:47
  SPEAKER_03: So, what's happening today, and what I call in the fourth intelligence revolution, is we're going even further to a point where now everyone is becoming a target of intelligence.
• 7:01
  SPEAKER_03: So we're sometimes see in the news that an insurance company or a bank has been hacked by China or Russia or the North Koreans.
• 7:09
  SPEAKER_03: So they're collecting information on regular people instead of just politicians and military officers and so forth, but also targeting regular people using information operations and things like that.
• 7:22
  SPEAKER_03: So think TikTok and that debate, for example.
• 7:26
  SPEAKER_03: At the same time, now anyone can do intelligence on their own.
• 7:31
  SPEAKER_03: So it's not just the CIA doing open source intelligence.
• 7:35
  SPEAKER_03: Regular people can do this.
• 7:37
  SPEAKER_03: And so there are organizations like Bellingcat, for example, which is a citizen journalism site that are effectively intelligence agencies where anyone can sign up and they can go and collect and analyze information and create intelligence products.
• 7:51
  SPEAKER_03: And many companies do this.
• 7:53
  SPEAKER_03: Um, so we're sort of entering a world now.
• 7:55
  SPEAKER_03: What's really new here is that intelligence is both targeting regular people, but also regular people can act like intelligence officers and use the tools that they have, up to and including commercial satellite imagery and so forth.
• 8:09
  SPEAKER_03: And that is where I see the revolution happening and perhaps what we can talk further about today.
• 8:15
  SPEAKER_00: Okay.
• 8:16
  SPEAKER_00: Fantastic.
• 8:16
  SPEAKER_00: Thank you so much.
• 8:17
  SPEAKER_00: Um, Ellie, I want to pick up on what Anthony mentioned about OSINT.
• 8:22
  SPEAKER_00: As he mentioned, it creates a much larger role for this open source intelligence.
• 8:27
  SPEAKER_00: Could you please explain for us what exactly is OSINT and what differentiates it from this traditional classified government intelligence?
• 8:37
  SPEAKER_02: Sure.
• 8:38
  SPEAKER_02: So uh open source intelligence really is using uh what we call PAI publicly available information.
• 8:47
  SPEAKER_02: Uh in other words, uh a member of the general public could reasonably be expected to gain access to the information through request observation or purchase back to commercial satellite imagery, right?
• 9:01
  SPEAKER_02: And then um that last category, we also buying content, buying information.
• 9:07
  SPEAKER_02: We call that commercially available information.
• 9:10
  SPEAKER_02: So open source information basically is uh or open source intelligence is we using publicly available information, commercially available information, it gets vetted, analyzed, and then we produce an answer to uh respond to an intelligence question.
• 9:30
  SPEAKER_02: So in the intelligence community, right, we deal with intelligence, national security type questions.
• 9:36
  SPEAKER_02: Um the term has become much broader, and so it's used in commercial due diligence investigations in cybersecurity, uh in a whole host of areas, citizen journalism.
• 9:49
  SPEAKER_02: The I guess the main distinction between open source intelligence and the and the other four intelligence discipline, human intelligence, signals intelligence, geospatial, and then measurement and signatures uh intelligence is that for it to be publicly available or commercially available information, um, somebody else had to go out collect that information for their own purposes, their own reasons.
• 10:18
  SPEAKER_02: And so we like to say open source intelligence is not a proactive collection activity in the sense that we're engaging with the target of collection, right?
• 10:33
  SPEAKER_02: So if you start asking somebody on the internet additional details, you've moved out of open source intelligence and now you're eliciting information, which is human.
• 10:44
  SPEAKER_02: So it's kind of a passive activity, a read-only type activity.
• 10:48
  SPEAKER_02: So if it involves hacking a website or, you know, recruiting sources where that's well outside of the um OSINT arena.
• 10:59
  SPEAKER_02: Uh and Anthony's point, uh, yeah, this goes back a long time.
• 11:04
  SPEAKER_02: Uh, we just highlighted on our LinkedIn page uh an article written in 2023 talking about George Washington's use of newspapers during the Revolutionary War.
• 11:16
  SPEAKER_02: Both sides did, and you know, it goes back way, way before that.
• 11:20
  SPEAKER_02: So that's a quick and dirty, basically uh uh using information that's already been collected and published by somebody else, um, which gets into some unique requirements for how we uh vet and validate that information, and we can chat about that more later.
• 11:38
  SPEAKER_00: Yeah, I I'd actually like to stay on that point just for a moment.
• 11:42
  SPEAKER_00: So there's all this information out there, and then you have people who are trained analysts, right?
• 11:48
  SPEAKER_00: And they're the ones who are really like the like the and correct me if I'm wrong, kind of the true OSINT operators, and then you have the general public with this democratization of information that is publicly or commercially available.
• 12:03
  SPEAKER_00: So how does that let's say, how does that division muddy the waters between you have what you know what the government's doing with its own OSINT analysts, but then there are all these private actors and citizen actors.
• 12:17
  SPEAKER_00: Um, are there any risks that you see in that in that balance or that dichotomy?
• 12:23
  SPEAKER_02: Yeah, so there's certainly risk in the sense that we have tradecraft, a way of doing business.
• 12:30
  SPEAKER_02: And and so there's some dangers associated.
• 12:34
  SPEAKER_02: In particular, we're very focused on ensuring that we're not giving away more information than we're collecting.
• 12:41
  SPEAKER_02: Uh, and so, uh, you know, if you're going after dangerous groups and uh you don't use good operational security, um, you can not only compromise what you're doing, but you know, in a in a war-time situation, in an active conflict, these folks, given the opportunity, will kill you, right?
• 13:03
  SPEAKER_02: And so that's critically important.
• 13:05
  SPEAKER_02: Or you may betray operational components that get somebody else killed.
• 13:11
  SPEAKER_02: Um, citizen journalism, they don't have that necessarily level of concern.
• 13:18
  SPEAKER_02: But if you're exposing something that a country doesn't want, I um would be very careful to go against some of the totalitarian governments exposing their data if I was John uh, you know, John Q public.
• 13:32
  SPEAKER_02: So that's critically important.
• 13:34
  SPEAKER_02: The other thing is there's some ethical considerations.
• 13:38
  SPEAKER_02: The uh Ukraine conflict, for example, early on we were very interested, and as were the Ukrainians, as to where the Russians had their mobile launchers, and you know, there's kind of a famous slash infamous photograph of a Russian tourist on a beach in Crimea, Crimea, wearing just inappropriately small speedos, and uh he's a he's a big dude, big hairy dude.
• 14:06
  SPEAKER_02: So, you know, it's kind of like uh but behind him you can see a mobile launcher that they were hiding, right?
• 14:14
  SPEAKER_02: And so that picture made the rounds and the Ukrainians targeted that location, and uh, and while it's a funny example, right?
• 14:25
  SPEAKER_02: You could get that person in a fair amount of trouble um inadvertently because somebody took a picture and maybe the Russians were very upset with the woman who took a picture of her husband in the Speedos on that beach, right?
• 14:39
  SPEAKER_02: And so we have to be very careful with what's disseminated.
• 14:45
  SPEAKER_00: Thank you both for setting the scene in this way.
• 14:48
  SPEAKER_00: I wanted to go now into just a couple of kind of real-world cases just to help us better understand the this changing nature of intelligence.
• 14:58
  SPEAKER_00: So, Anthony, I wanted to start with something that you mentioned briefly, which was TikTok.
• 15:02
  SPEAKER_00: You know, off you know, the fight over TikTok has been in the news.
• 15:05
  SPEAKER_00: I'd love to just get your take on how the ownership of this platform matters from an Intel perspective and use as well, and what that says about the balancing of citizen rights of access to information over what might be perceived as a national security threat.
• 15:22
  SPEAKER_03: Yeah, it's a great question.
• 15:24
  SPEAKER_03: Well, a lot of people think about TikTok in terms of the information on it being stolen or used by a foreign power.
• 15:34
  SPEAKER_03: So TikTok initially was owned by China.
• 15:37
  SPEAKER_03: Um actually they merged an American company musically to create it, but it was owned by China.
• 15:44
  SPEAKER_03: The servers holding all of their data were located in mainland China.
• 15:50
  SPEAKER_03: And so an American user on TikTok uh making videos or entering their information or even swiping on what they want to look at, all of that information was being transferred to China.
• 16:04
  SPEAKER_03: And even in cases where they weren't supposed to transfer that information due to other American legal reasons and including um miners' information.
• 16:15
  SPEAKER_03: And so this is well documented, and there have been various lawsuits on this.
• 16:19
  SPEAKER_03: So at one level, the the problem with TikTok is that China was taking American citizens' data, bringing it back onto China's mainland, and then we had no control over that data from there.
• 16:32
  SPEAKER_03: Data like that can be used to target Americans.
• 16:35
  SPEAKER_03: It could be used to suss out who's an intelligence officer or who might become an intelligence officer or who has a medical condition that would make them susceptible to recruitment or bribery or something like this.
• 16:50
  SPEAKER_03: Um there is a real issue there when you have any really any technology, but especially a kind of an information technology owned and maintained by an authoritarian state and one that is an adversary to us at the end of the day.
• 17:04
  SPEAKER_03: There's something else, though, going on with TikTok that's even more dangerous, which is that that information can be used to change or to influence people's views.
• 17:14
  SPEAKER_03: So this isn't about just collecting information, it's about doing something.
• 17:19
  SPEAKER_03: Um we call this covert influence, um, when you try to influence a population covertly without acknowledging that you're doing that.
• 17:28
  SPEAKER_03: So it's different than advertising.
• 17:30
  SPEAKER_03: Advertising, you're acknowledging, you're saying, I'm trying to sell something.
• 17:33
  SPEAKER_03: Covert influence, you're sort of pretending as if you're not doing it and you're influencing a population.
• 17:40
  SPEAKER_03: And when an authoritarian country like China owns this platform, they can do that.
• 17:46
  SPEAKER_03: There is a lot of evidence that they've done this.
• 17:48
  SPEAKER_03: You can look statistically to see, for example, that the there, and there have been studies um at Rucker's and some other universities showing that the longer a user is on TikTok, the more benevolent view of the Chinese Communist Party's human rights record they will have.
• 18:08
  SPEAKER_03: So there's evidence showing that it will work.
• 18:10
  SPEAKER_03: There's also statistical evidence showing that the that on TikTok versus on an American-owned platform like Instagram or something like this, more information will be censored about Tiananmen Square and the Uyghurs human rights record and various issues that concern China.
• 18:29
  SPEAKER_03: So you can see that they are actually changing what is available and it is working to change American citizens' views.
• 18:39
  SPEAKER_03: So there's proof that there is, and this is why this was such a big issue and why eventually the US government had to transfer it.
• 18:47
  SPEAKER_03: So it becomes a tool not just of intelligence collection with data, but actual intelligence information operations against people.
• 18:58
  SPEAKER_00: Thank you for clarifying that.
• 19:00
  SPEAKER_00: Another area I'd like to dig into with both of you is AI.
• 19:04
  SPEAKER_00: It's impossible to avoid now.
• 19:06
  SPEAKER_00: Elliot, you've specifically identified the growth of AI-generated content on social media and other publicly available sources as a trade crap problem for OSINT practitioners.
• 19:18
  SPEAKER_00: So in plain English, what changes when an analyst can no longer assume that, let's say, a photograph, a voice, a video, a document, or an event, or even the person posting it is real?
• 19:31
  SPEAKER_00: And if possible, could you give us a concrete example on how like synthetic material might enter an OSINT investigation?
• 19:39
  SPEAKER_02: Yeah.
• 19:39
  SPEAKER_02: So uh, you know, OSINT has always had this issue of having to validate because uh we don't own the means of collection.
• 19:48
  SPEAKER_02: So with other disciplines, right, we own the satellite or the plane that takes the picture, right?
• 19:56
  SPEAKER_02: We don't own the human source, but we have a relationship with the human source.
• 20:01
  SPEAKER_02: Um, and you know, we're listening in with our equipment.
• 20:06
  SPEAKER_02: So we don't have to question whether that content was really collected with open source, because again, we're sort of secondhand users of data that somebody else collected.
• 20:18
  SPEAKER_02: So that's always been an issue.
• 20:20
  SPEAKER_02: Is that really in the past it was more recycling of photographs, right?
• 20:27
  SPEAKER_02: So this was a photograph in Afghanistan three years ago, and we're gonna show it today, and we're gonna change it a little bit.
• 20:34
  SPEAKER_02: So it looks like a US atrocity.
• 20:38
  SPEAKER_02: And uh and so uh we've always had to do that vetting and analysis of photographs and video and whatnot.
• 20:45
  SPEAKER_02: What's changed now is just that all of that can be orchestrated using AI.
• 20:52
  SPEAKER_02: So those photographs can so we're moving from, you know, is this has this photograph been manipulated or used from someone else or somewhere else to now is this even a real photograph?
• 21:06
  SPEAKER_02: Uh and so we have tradecraft, we have methods to ascertain that and so and we see that consistently.
• 21:15
  SPEAKER_02: You see that across the board.
• 21:17
  SPEAKER_02: Uh the Chinese, the Russians, uh, the North Koreans uh are in some cases generating content out of whole cloth and posting it.
• 21:27
  SPEAKER_02: And so it has made our life a little more difficult.
• 21:31
  SPEAKER_02: If you're an open source officer in the intelligence community, you're having to rely more on that uh kind of technical expertise.
• 21:41
  SPEAKER_02: But you know, with a challenge comes an opportunity, and the good side of that is it gives us some insight into intent, right?
• 21:50
  SPEAKER_02: So why is the adversary generating this particular uh image, right?
• 21:56
  SPEAKER_02: Uh rather than well, maybe because they're Latest generation fighter plane really can't do what it was supposed to do.
• 22:05
  SPEAKER_02: And so, you know, now we've got to augment the reality to try to sell the plane or to sell the concept that Country X is on par with the US.
• 22:15
  SPEAKER_00: That's fascinating.
• 22:16
  SPEAKER_00: Anthony, I know you've also written a lot about AI and have done a lot of forecasting about the intersection of AI and the Intel space.
• 22:24
  SPEAKER_00: And in particular, if machines become targets of other machines' intelligence operations.
• 22:30
  SPEAKER_00: So would you mind just digging into what you're looking ahead to when it comes to AI intelligence and what that might mean for an allied government trying to decide whether a piece of commercial satellite or OSINT data has been deliberately poisoned even before it even reaches analysts?
• 22:51
  SPEAKER_03: I think it's helpful to think about AI in the intelligence community like we think about AI in all industries right now.
• 22:59
  SPEAKER_03: You know, we're we're all sort of seeing how Claude or ChatGPT can be used to draft legal documents.
• 23:07
  SPEAKER_03: And so lawyers are wondering what to do with AI, or it can be used for accounting or for consulting or various forms of business, right?
• 23:15
  SPEAKER_03: Even education and so forth.
• 23:17
  SPEAKER_03: The same thing is happening in the intelligence community today.
• 23:21
  SPEAKER_03: And AI is being used kind of across the entire system of intelligence.
• 23:27
  SPEAKER_03: We think about it in terms of the intelligence cycle where you're collecting information, you're analyzing it, you're disseminating it, you're planning to collect again and so forth.
• 23:36
  SPEAKER_03: Well, in each of those steps, you can use AI.
• 23:39
  SPEAKER_03: And so, in particular, one area that has had a lot of growth recently in AI intelligence is analysis.
• 23:49
  SPEAKER_03: Because if you think about it, if you know, for all of us that have gotten on Claude or ChatGPT, like it's actually pretty good at analysis, right?
• 23:57
  SPEAKER_03: It can take in a lot of different data feeds, it can search the entire internet in effect and answer a question for you.
• 24:05
  SPEAKER_03: The the the goal in the intelligence community is to kind of do the same thing.
• 24:10
  SPEAKER_03: Um and you're already seeing agencies so that doing this.
• 24:13
  SPEAKER_03: So the the director of NGA, my old agency, announced that they had were now creating intelligence reports that he said no human hand had touched, so we're AI generated.
• 24:25
  SPEAKER_03: Um the deputy director of the CIA, Michael Ellis, recently announced that they were generating intelligence reports using AI.
• 24:33
  SPEAKER_03: So this is kind of happening.
• 24:35
  SPEAKER_03: And where I think this is going is that the entire process of intelligence will in short time be it'd be possible to do that using AI at every step.
• 24:46
  SPEAKER_03: And that in fact, some of these intelligence reports will just be passed to other AI systems, maybe to a drone or to some other AI operational system.
• 24:56
  SPEAKER_03: Um, but what's interesting is that our adversaries are doing the same thing.
• 25:01
  SPEAKER_03: So China is out there also using AI to perform intelligence, to collect information and analyze it and so forth.
• 25:10
  SPEAKER_03: And we're kind of going to enter this world where we're gonna have their AI systems performing intelligence on our AI systems and vice versa.
• 25:19
  SPEAKER_03: We're having AI systems collecting information about other AI systems and analyzing it and targeting and so forth.
• 25:26
  SPEAKER_03: Um, this sounds a lot like science fiction, but we kind of already see it in one industry.
• 25:31
  SPEAKER_03: We we see it in finance, where you know, at hedge funds and so forth, there's a sort of a branch of um public market hedge fund trading called quantitative or quant trading, where you have an AI system in effect collecting information about markets, figuring out what to trade, and then performing those trades.
• 25:50
  SPEAKER_03: And there are other quant funds out there doing the same thing.
• 25:54
  SPEAKER_03: And now these two sys these two different competing AI systems are trading against each other at a speed and a scale that no human being can kind of see.
• 26:04
  SPEAKER_03: And so that it's already kind of happening, and I think the same thing is going to happen in intelligence.
• 26:10
  SPEAKER_03: And so for the open source world, this is and and for intelligence in in general, this is this is kind of a new world that is rife with all sorts of risks, as you might imagine, right?
• 26:23
  SPEAKER_03: But first of all, is it collecting the right information?
• 26:26
  SPEAKER_03: Is it maybe collecting faked information, like Elliot was just saying, it's true.
• 26:30
  SPEAKER_03: The anything, any video, audio, or writing can be faked at this point by AI.
• 26:36
  SPEAKER_03: So how do we know it's real?
• 26:37
  SPEAKER_03: So, and we could create an intelligence report purely built on fake information.
• 26:43
  SPEAKER_03: Um, and then that obviously would be very dangerous, could lead to loss of human life, in fact, if uh it was used for targeting, for example.
• 26:52
  SPEAKER_03: Um, so lots of lots of sort of practical and ethical dangers in that.
• 26:57
  SPEAKER_03: It also means, by the way, and getting back to where I started, that people can kind of do this too.
• 27:03
  SPEAKER_03: Um, you know, there are there are professional intelligence officers, but remember, um, the AI systems are very good at doing this.
• 27:13
  SPEAKER_03: And most of this AI tech is not created by Lockheed Martin directly for, you know, a government.
• 27:19
  SPEAKER_03: I don't mean to pick on Lockheed.
• 27:21
  SPEAKER_03: It could be Raytheon or anybody, right?
• 27:23
  SPEAKER_03: Palantir, some big contracting company.
• 27:26
  SPEAKER_03: Most of the AI technology that's really truly cutting edge is available to all of us.
• 27:31
  SPEAKER_03: And so we could all go and use this technology to generate, to collect and analyze and generate our own intelligence reports.
• 27:39
  SPEAKER_03: We can do the same thing as these intelligence agencies.
• 27:44
  SPEAKER_00: Wow.
• 27:45
  SPEAKER_00: That's a lot to take in.
• 27:46
  SPEAKER_00: Um continuing this look ahead, Elliot, Anthony just brought up this, you know, this interesting point about speed and this distribution of intelligence.
• 27:58
  SPEAKER_00: So as intelligence collection is now distributed more widely among governments, among companies, machines, and citizens, what do you think must change in order to ensure trust, credibility, and effectiveness?
• 28:10
  SPEAKER_00: And then just to layer on that, as Anthony alluded to, there's a speed issue now that probably always existed, but everything just seems to be going faster and faster.
• 28:21
  SPEAKER_00: And so how do you, in a competitive geopolitical environment, ensure the verification with the speed question?
• 28:30
  SPEAKER_02: So, yeah.
• 28:31
  SPEAKER_02: So let me let me take the speed issue first.
• 28:34
  SPEAKER_02: Um, you know, for many years I've been doing OSINT since the early 1990s.
• 28:41
  SPEAKER_02: And in those days, it was trying to find the needle in the haystack.
• 28:46
  SPEAKER_02: Uh, that was a problem.
• 28:48
  SPEAKER_02: Uh, that um it wasn't so much that we couldn't collect the data as that there was so much data that we couldn't sift through it.
• 28:57
  SPEAKER_02: Uh, and so we've moved from trying to find that needle in a haystack to just saying, just find the haystack, right?
• 29:06
  SPEAKER_02: And then we'll we have automated tools that'll find the needle, right?
• 29:10
  SPEAKER_02: And with AI, now we're saying just grab the whole field, right?
• 29:14
  SPEAKER_02: And we'll worry about those haystacks later.
• 29:16
  SPEAKER_02: And if you think about those haystacks as kind of the knowledge repositories, the underlying data that AI is using, those large language models, what small language models, uh, so those haystacks become more important more important.
• 29:34
  SPEAKER_02: Uh, and so there's unique holdings that only the US government has of technical data that may be unclassified, but it's not publicly available per se, the the analysis of that data.
• 29:49
  SPEAKER_02: So the speed issue, what's helped us tremendously is we have the ability to go out, grab huge sums of data and uh and process it very quickly.
• 30:01
  SPEAKER_02: So that's a big plus for the OSINT arena.
• 30:05
  SPEAKER_02: Um, and in fact, you know, all of the work that the intelligence community is doing with AI, all of that we in the OSINT discipline were the early adopters of that.
• 30:17
  SPEAKER_02: And uh the OSINT community was the first to establish standards for how do we cite that data?
• 30:25
  SPEAKER_02: How do we cite publicly available information, and then how do we vet it?
• 30:31
  SPEAKER_02: And then, you know, how do we annotate when we're using AI, right?
• 30:35
  SPEAKER_02: Because transparency is critically important.
• 30:38
  SPEAKER_02: That war fighter or that policymaker should know that AI was part of the process and exactly how it was part of the process.
• 30:46
  SPEAKER_02: And so um from that perspective, that helps us tremendously because we've always been underfunded and understaffed in open source.
• 30:56
  SPEAKER_02: Uh, and so we now can move our open source officers to that higher level function, that value added that's critical.
• 31:06
  SPEAKER_02: Back to what has to change, right?
• 31:09
  SPEAKER_02: So we have to be extremely careful with those haystacks, right?
• 31:15
  SPEAKER_02: We have to ensure the integrity of that data, because nothing would be more advantageous than manipulating the data your adversary is using to do its AI work.
• 31:29
  SPEAKER_02: So that's critically important.
• 31:31
  SPEAKER_02: And so not only being very careful in how we buy the data, but uh, but then also in how it's collected.
• 31:39
  SPEAKER_02: If we're using automated collection tools, I had a conversation with somebody who were talking about you know the operational security around open source intelligence and how we carefully collect the data without other people realizing we're collecting it.
• 31:54
  SPEAKER_02: And they said, Oh, we don't have to worry about that anymore, we'll just use AI to collect the data.
• 31:59
  SPEAKER_02: And uh I'm like, that's even worse, right?
• 32:02
  SPEAKER_02: The problem is more acute in that instance because you're doing it once and then you're gonna share it broadly with a bunch of people.
• 32:11
  SPEAKER_02: And so if the adversary figures out how your AI is collecting the data, it's very easy to manipulate.
• 32:19
  SPEAKER_02: In the past, it was difficult to manipulate open sources because you'd have to try to manipulate all of the sources of information.
• 32:27
  SPEAKER_02: Nowadays, with AI, it's much easier to seed the information in a thousand places in a matter of seconds.
• 32:36
  SPEAKER_00: Anthony, also one thing that we haven't really touched on directly is the impact on the private sector and sort of what the economic impacts are.
• 32:47
  SPEAKER_00: So how does the private sector factor in?
• 32:50
  SPEAKER_00: Is today's espionage and let's say espionage's connection with the private sector, is it different from say like France's suitcase spying on US companies in the 70s and 80s?
• 33:01
  SPEAKER_00: Like how, in short, like how does Intel and the democratization of intelligence matter to the economy today?
• 33:10
  SPEAKER_03: Yeah, great question.
• 33:11
  SPEAKER_03: Well, look, you know, again, if you look back historically, there's always been some level of economic espionage.
• 33:20
  SPEAKER_03: In fact, we in America sent effectively industrial spies to England and Scotland in the 1800s to kind of steal IP and uh and brought it back here.
• 33:33
  SPEAKER_03: So there's a long history on all sides.
• 33:37
  SPEAKER_03: And and and and and you're right, that's a great example of kind of French industrial espionage in the 70s.
• 33:43
  SPEAKER_03: And the the Soviets um tried to do it to us during the Cold War.
• 33:48
  SPEAKER_03: What's happening today is that it's it's really scaled up into something entirely different than we've ever really seen.
• 33:57
  SPEAKER_03: And part of what's driven that is that the competition between the US and China in particular is very much an economic competition, as much as it is a military um or a political competition globally.
• 34:11
  SPEAKER_03: Um and China has, and again, this is fairly well established in various sources that China's kind of had a very significant espionage campaign, economic espionage campaign over the last 20 years, um, targeting US businesses.
• 34:27
  SPEAKER_03: Um and that is that is grown in a way we've never really seen.
• 34:32
  SPEAKER_03: And they're using cyber attacks to quite literally hack into company um databases and steel IP.
• 34:40
  SPEAKER_03: They're using individual people, human sources to work at companies or recruit somebody who works at companies in America to steal intellectual property.
• 34:50
  SPEAKER_03: So this is kind of new and related to what I said before, which is that the scope of intelligence has really expanded beyond the world of political and military issues into the economy.
• 35:02
  SPEAKER_03: And that includes companies that you know are making kind of regular stuff, like not just don't think in your mind aerospace companies making spaceships.
• 35:13
  SPEAKER_03: You know, this might be consumer technologies and really anything manufactured is potentially uh part of the software companies.
• 35:23
  SPEAKER_03: Um, there's uh plenty of warnings, for example, to um Silicon Valley startups by the FBI and other law enforcement agencies that they have to be very careful about insider threats from China.
• 35:35
  SPEAKER_03: So companies, the economic domain has become part of the great game of espionage, as it were.
• 35:42
  SPEAKER_03: At the same time, companies now have access to tools, to intelligence tools to help them.
• 35:52
  SPEAKER_03: They can use very sophisticated um software systems for countering insider threats.
• 35:59
  SPEAKER_03: Um, they can tap into all sorts of cybersecurity measures that um, you know, are many of these companies are quite literally run by former NSA employees, right?
• 36:12
  SPEAKER_03: Like they're tapping right back into that knowledge base.
• 36:15
  SPEAKER_03: Also, what we're seeing, you know, what's been interesting in the last several months is seeing how geopolitical activities are affecting the bottom lines of companies, right?
• 36:25
  SPEAKER_03: So you you look at what's happened in Iran with the closure of the Strait of Hermuz, Iran was obviously aware that closing the Strait of Hermuz would have economic implications and sort of, and that would harm individual companies and the oil industry and create inflation for consumers that would have an effect on midterms and things like this.
• 36:47
  SPEAKER_03: They were aware of that.
• 36:49
  SPEAKER_03: And so the economy has become part of that war, and they're using information operations to play that up.
• 36:56
  SPEAKER_03: And now you as a company have to discern what Iran is doing, what's going on in geopolitics, whether they're using information operations to release information to purposely make it look like this is a worse situation and so forth.
• 37:12
  SPEAKER_03: So companies are now part of the battleground, but also have the tools and the need to act like intelligence agencies in a sense to defend themselves.
• 37:23
  SPEAKER_00: Great.
• 37:24
  SPEAKER_00: Thank you.
• 37:24
  SPEAKER_00: Um I'm about to turn to audience questions, um, but I just want to end this moderated conversation with a really quick signal check, basically a lightning round based on our discussion so far.
• 37:35
  SPEAKER_00: So, really just looking for the headline answer in one or two sentences, and don't worry if it repeats something that you already covered.
• 37:41
  SPEAKER_00: We really just want to highlight the takeaways.
• 37:43
  SPEAKER_00: So, kind of going to your point, Anthony, right now on the geopolitical competition.
• 37:48
  SPEAKER_00: I'll start with you and then go to Elliot.
• 37:50
  SPEAKER_00: Right now, who is adapting faster to this new intelligence environment?
• 37:54
  SPEAKER_00: Is it the US and its allies or their adversaries?
• 38:00
  SPEAKER_03: Unfortunately, I think China might be adapting faster.
• 38:05
  SPEAKER_02: I would concur with that assessment.
• 38:07
  SPEAKER_00: Ah, thank you.
• 38:08
  SPEAKER_00: Elliot, this one's for you.
• 38:09
  SPEAKER_00: Should OSINT remain a distinct professional intelligence discipline or should it become a basic competency that's expected of every intelligence analyst?
• 38:18
  SPEAKER_02: So some degree of competency is essential because all source analysts going out on the internet themselves results in what I like to call surfing fratricide, where we uh betray uh our interest in a particular topic, or somebody untrained going out may discover something that the adversary doesn't know is of value, and then with poor tradecraft, they demonstrate.
• 38:58
  SPEAKER_02: Are speaking from a 1990s concept of what OSINT is, it is so highly specialized nowadays and so sophisticated that um uh there's no way that an all-source analyst can do that.
• 39:16
  SPEAKER_02: And frankly, we pump information from human, from SIGIN, from Mazin, from Geoint to that all-source analyst.
• 39:25
  SPEAKER_02: They deserve a dedicated, super high quality pipeline of validated information.
• 39:32
  SPEAKER_02: And so, yes, it should be a standalone, it is a standalone discipline.
• 39:37
  SPEAKER_02: It should be a standalone career field with adequate resources.
• 39:42
  SPEAKER_00: Thank you.
• 39:43
  SPEAKER_00: Anthony, what's one intelligence responsibility that the government should never outsource?
• 39:51
  SPEAKER_03: Oh never outsource.
• 39:53
  SPEAKER_03: I think that um we want to keep counterintelligence internal to the government.
• 40:00
  SPEAKER_03: Uh, just literally the incorporation of somebody just adds another attack surface and and is probably more harmful than it's worth.
• 40:09
  SPEAKER_00: All right.
• 40:10
  SPEAKER_00: And for both of you, what is one concrete change you would want to see by this time next year that would show that the US and its allies are adapting to this new Intel environment?
• 40:20
  SPEAKER_00: So, Elliot, I'll start with you and then Anthony.
• 40:23
  SPEAKER_02: Uh, I'd love for the intelligence community to do a GSA schedule for OSINT.
• 40:30
  SPEAKER_00: Can you explain a GSA schedule?
• 40:33
  SPEAKER_00: Yeah, so it's for us kind of government.
• 40:36
  SPEAKER_02: Yeah, the US government purchases lots of things.
• 40:39
  SPEAKER_02: Uh, and um they have a simplified contract vehicle through the General Services Administration, the GSA, and and it's basically a schedule of things you can buy.
• 40:52
  SPEAKER_02: And it's a pre-negotiated contract, and the rates are publicly available, and I think it's a five-year contract, so you know in year three what an hour of uh you know, uh Russian linguist is gonna cost you or this particular tool or that software package, and so rather than the government the 18 agencies buying the contra with 18 different contracts, it would be great to just have that one pre-negotiated contract.
• 41:29
  SPEAKER_00: Great, Anthony, what about you?
• 41:32
  SPEAKER_03: Um intelligence agencies share information with each other, so uh including across countries.
• 41:39
  SPEAKER_03: So kind of the biggest one for America is called Five Eyes.
• 41:42
  SPEAKER_03: We share information with the the British and the Canadians and some other countries, English-speaking countries.
• 41:48
  SPEAKER_03: I would love to see something like that two-way intelligence sharing with an American industry.
• 41:55
  SPEAKER_03: Um, so with finance, for example, um, focused, uh, think about it as a public-private partnership where we're sharing information, not not law enforcement, I'm talking intelligence to help help each side.
• 42:08
  SPEAKER_00: Great.
• 42:08
  SPEAKER_00: Thank you.
• 42:09
  SPEAKER_00: Given the various types of intelligence that have been discussed, how would you rank the importance of human intelligence and if known, what percentage does it contribute to overall intelligence?
• 42:17
  SPEAKER_00: And I would also add, is there a danger of the government not adapting in the AI world such that are they going to be able to adequately fund traditional intelligence collection?
• 42:30
  SPEAKER_03: Yeah, good question.
• 42:31
  SPEAKER_03: I mean, I'm a former human intelligence officer, a case officer.
• 42:35
  SPEAKER_03: So I think there was a fear that with uh more technology and AI, you'd have less human intelligence and you would rely more on AI and things to collect.
• 42:46
  SPEAKER_03: I actually think it's the opposite.
• 42:48
  SPEAKER_03: And I think as there's more AI in the world, you actually need more human intelligence because we, as an all people, are going to stop trusting anything we see online because anything can be faked.
• 43:05
  SPEAKER_03: So anything that's sent to you as an email or that you hear as a video and so forth, you just won't trust it at all.
• 43:12
  SPEAKER_03: And we're all going to kind of turn inward in that sense, especially people working in worlds like national security.
• 43:20
  SPEAKER_03: And so we're going to have to have people who specialize in connecting with other human beings in person, who speak that language, know that culture, and can go meet other people and recruit them and get them to provide secrets.
• 43:34
  SPEAKER_03: And you can have all the cybersecurity in the world that you want to protect a secret, but still there's a human being who goes into that office and who can extract the secrets for you.
• 43:43
  SPEAKER_03: And uh that those secrets, by the way, might be on the design of an AI system, right?
• 43:48
  SPEAKER_03: You can great, you have amazing cybersecurity on your world class AI model.
• 43:56
  SPEAKER_03: Um, let's see if a human intelligence officer can recruit the designer.
• 44:00
  SPEAKER_03: of that AI model to help our cause.
• 44:03
  SPEAKER_03: So I think that's where this world is going.
• 44:04
  SPEAKER_03: So if anything, I think what we need to do is actually be recruiting more human officers, not less, and growing the importance of that field.
• 44:13
  SPEAKER_02: Courtney, can I jump in?
• 44:15
  SPEAKER_02: Please.
• 44:16
  SPEAKER_02: I don't I think we've done ourselves a lot of disservice with pitting one intelligence discipline against another.
• 44:23
  SPEAKER_02: Right.
• 44:23
  SPEAKER_02: Uh it's kind of like if you're building a house, it's the argument of do we need more plumbers than electricians?
• 44:31
  SPEAKER_02: No, we need better plumbers.
• 44:33
  SPEAKER_02: We need better electrician, right?
• 44:35
  SPEAKER_02: We want more exquisite of each one of those.
• 44:39
  SPEAKER_02: Right.
• 44:39
  SPEAKER_02: And so they are all critical to build a house.
• 44:42
  SPEAKER_02: You can't build a house without masons or at least one that will last for long.
• 44:47
  SPEAKER_02: Uh and so you know each of those intelligence trades are critically important.
• 44:54
  SPEAKER_02: And so it's not a question of, oh well, you know, we'll we'll we'll ramp up OSIN and then we can start doing away with the others.
• 45:02
  SPEAKER_02: That will never happen.
• 45:04
  SPEAKER_02: And it shouldn't because it would result in poor information, poor intelligence for our consumers, whether they're law enforcement or uh law enforcement or operational commanders or policymakers.
• 45:19
  SPEAKER_00: Anthony, what is the most significant avenue that malign foreign actors like Russia and China would use for espionage, which you believe is being overlooked and how can we rectify this oversight?
• 45:33
  SPEAKER_03: I think that there I'm actually writing an op-ed um that is going to be coming out pretty soon on the topic of LLM grooming which is just a horrifying name for something.
• 45:46
  SPEAKER_03: And the the idea behind this kind of attack is that you generate a lot of news material that isn't targeted at people.
• 45:56
  SPEAKER_03: It's targeted at the AI systems.
• 45:59
  SPEAKER_03: And then that that means that when that when a person uses that AI system at in in the end, they're going to they're going to see that propaganda that has been targeted at the AI system.
• 46:12
  SPEAKER_03: So there's um an organization called the Provda Network out of Russia that does this um where they they literally have hundreds and hundreds of websites and they have they publish million literally millions of documents these are fed you know these are in the internet out there.
• 46:28
  SPEAKER_03: And then if I'm on Claude and I use Claude Claude is going to pick some of that up and when I ask it a question especially a question having to where there's not a lot of other information on the internet about it say I'm asking about some obscure topic in Russia um it will Claude will think that this is just a regular news site and it will present that information to me.
• 46:52
  SPEAKER_03: And the problem is that we are all beginning to kind of trust these AI systems a lot like Claude and and ChatGBT and so forth.
• 47:01
  SPEAKER_03: They're very convincing.
• 47:02
  SPEAKER_03: They you know we we sort of believe them for various reasons.
• 47:06
  SPEAKER_03: And um so now I'm going to believe what this propaganda says and I'm I could ask Claude for access to the ultimate source material but often don't um and so now it becomes this trajectory for propaganda that didn't exist even two or three years ago and now is I think going to be the main means by which like we are subjected to propaganda and Russia is doing it, China is doing it.
• 47:33
  SPEAKER_03: This is becoming um a kind of standard standard thing and I think um we don't have a great fix for that right now either technically or law enforcement wise.
• 47:45
  SPEAKER_00: Interesting.
• 47:45
  SPEAKER_00: Yes it's like it's like poisoning the upstream with a charmingly confident interlocutor.
• 47:50
  SPEAKER_02: Elliot what has been the biggest surprise and how intelligence has changed things to OSINT tools so I think the biggest surprise is just how pervasive OSINT has become as the um enabler and precursor to the other ints and so it's being used you know in Geoint uh commercial imagery right was a very early instance of that and we consume a very significant amount of commercial imagery because it helps us with you know making uh that data available to other uh you know coalition partners and whatnot so it's a it's a great way to get information out without with again keeping our precious high highly capable systems out of view but uh you know OSIN is is being used across all the ints it's critical so you know one of the biggest consumers of PAI and CAI is NSA and so all of the ints are using it and uh as a starting off point because if we're targeting people why would we begin to try to recruit a source without first exhausting what's publicly available about that individual and we have all sorts of platforms nowadays that can be you know that'll give us all sorts of data right this thing that we all have and love and can't live without those are all a persistent collection platform right and so this thing pings a cell tower many times a second and your geolocational data is available your car collects everywhere you go that data is commercially available if your car doesn't collect it you know you have companies out there with automated license plate readers that not only capture your license plate but they'll capture the bumper stickers on your vehicle and so all of that is critically important.
• 50:03
  SPEAKER_00: So I just want to say thank you to both of you this is fascinating we're both visionaries um and I really appreciate your taking the time to be here today.
• 50:12
  SPEAKER_00: With that, Elliot, Anthony, thank you this was fascinating um we really appreciate it and um I wish you all well thanks so much for having me.
• 50:23
  SPEAKER_00: Thank you.
• 50:24
  SPEAKER_00: Bye-bye.
• 50:25
  SPEAKER_01: Take care of Global Insights you can subscribe to Global Insights on podcasts, Spotify or wherever you get your audio for more analysis, events and ways to connect with our community, click on the link in the description.
• 50:52
  SPEAKER_01: See you next time