Intelligence in a Transparent World

Human operations, once the bread and butter of spy work, changed dramatically in the last two decades, thanks to a proliferation of "smart city" technologies and biometric identity data.⁵ Back in 2010, the Emirati intelligence services were able to quickly identify the members of a Mossad operation that assassinated a senior member of Hamas in Dubai. Using surveillance camera footage, travel records, and phone records, they identified the Mossad operatives responsible for the attack within a month.⁶ Today, with AI-enabled facial recognition and Chinese companies selling security systems across the globe, it is too easy to connect dots and unravel an entire intelligence operation. Starting with an image of a suspected case officer meeting with an asset, an enterprising intelligence service can track that case officer's movements across the world over the last 10 years, using AI to identify everywhere they have been seen and whom they have been seen with. In 2018, a senior technology officer at the CIA said that in many places, "the level of surveillance was so mature that local security services no longer needed to follow the agency's officers in order to know where they were."⁷ Biometric passports make traveling under an assumed identity far more difficult, and false identities seem paper thin with no decades-long social media history to back them up. Plus, any border guard has the ability to fact-check backstories instantly. As The Economist points out: "A spy can be instantly quizzed on their assumed identity's childhood route to school by an enterprising immigration officer using Google Maps."⁸

OSINT has the potential to fill at least some of the gaps left by more challenging HUMINT. A multitude of industries have decided that data is the new oil and are mining every available source to create massive, useful datasets. According to the Office of the Director of National Intelligence's (ODNI) senior advisory group on commercially available information (CAI), "CAI includes information on nearly everyone that is of a type and level of sensitivity that historically could have been obtained, if at all, only through targeted (and predicated) collection, and that could be used to cause harm to an individual's reputation, emotional well-being, or physical safety."⁹ Beyond information on people, governments can obtain data on the health of shipyards based on soundscapes, the movements of submarines based on sonar designed to locate fish, or the location of tanks and troops based on commercial space assets.

Critical to open-source work will be recognizing the potential pitfalls of this relatively new discipline. First, the ubiquity of data means it can be selected or manipulated to fit nearly any narrative. Second, in an era where data is power, democracies must walk a tight corridor between harvesting information and protecting the rights of citizens. Finally, for every Bellingcat exposure of nefarious action, there are likely a handful of crises averted because of exquisite, highly classified intelligence collection. OSINT should be additive to the intelligence picture, even serving as the intelligence of first resort, but it cannot fully replace clandestine collection.

Lessons from Modern Wars

Ukraine's and Israel's recent conflicts have much to teach about the power of intelligence and where the discipline is headed. The conflict in Ukraine has been revolutionary on two fronts: First, it has been a truly open-source war, with crowd-sourced intelligence work making both a tactical and a strategic difference. Publicly and commercially available data has been pivotal to widespread sharing at a high level among allies and on a tactical level between units in the field. Second, the Biden administration's decision to declassify intelligence strongly indicating that Russia was about to invade teaches twin lessons-calling out Moscow's plans did not deter Russia from invading, but it did help pre-bunk ridiculous narratives and galvanize allies to assist Kyiv in blunting the Russian offensive.

Conversely, students of the practice of intelligence will study the tragedy of October 7, 2023, for decades. As a counterpoint, they will study Israel's astonishing victory over Hezbollah in the ensuing year, in which Israel systematically dismantled the group's fighting apparatus, for the opposite reason. Israel had all the information it needed to identify and preempt the Hamas attack, but cognitive bias prevented action. With Hezbollah, on the other hand, Israel took the threat from the group seriously and created in-depth, multiyear plans to strike, with devastating results when it eventually pulled the trigger.¹⁰ The failure-success juxtaposition of Gaza and Lebanon shows that a rigid mindset trumps even the most sophisticated intelligence, but the combination of detailed intelligence work and persistent attention to a threat can devastate even a talented adversary.

Ukraine: The First Open-Source War

Ukraine is the first truly open-source war. According to General Jim Hockenhull, commander of the United Kingdom's strategic command, OSINT has been instrumental in providing Ukrainian commanders with anticipatory intelligence.¹¹ Commercial satellite imagery, tech data, and social media helped expose Russian deployments well ahead of the February 2022 invasion. A Russian submarine commander reportedly was killed after logging his daily run on the fitness tracking platform Strava.¹²

Every citizen with a cellphone became a sensor, taking videos and photos of Russian troop movements. At first, they uploaded the geotagged images to social media. Then Kyiv adapted the Diaa app, originally designed to help Ukrainians access social services, to create the e-Enemy platform.¹³ By one estimate, 260,000 Ukrainians reported Russian locations to the app in the first month of the invasion.¹⁴ Ukraine's Security Service also welcomes reports of sightings of "suspicious" activity via a Telegram chat function called @stop_Russian_War_bot.¹⁵ Stories abound of Ukrainian commanders needing to know what was happening at a certain location, finding a business on Google Maps that was near that location, then calling to ask the proprietor to look outside and report what they saw:

"We open a Google map, see a store, see its phone number, and dial it," said Shevchuk, who described a typical conversation: "Good evening, we are from Ukraine! Do you have any Katsaps [Ukrainian slur for Russians] in the village?" "Yes." "Where?" "Behind Grandma Hanna's house." "What house does Grandma Hanna have?" "Well, everyone knows her!" "So you talk to people a little bit and realize where everything is," Shevchuk added.¹⁶

Classic honey traps have evolved for the online space. Defense Mirror reported that a Ukrainian woman used several Tinder profiles to collect information on more than 70 Russian soldiers, which she promptly passed to Ukrainian troops to help with strikes.¹⁷ Similarly, MI-6 reportedly used Grindr to find Russian troops.¹⁸

Commercially available intelligence services have been a game changer in Ukraine. Kyiv has leaned into a relationship with space technology company Maxar, which provides fairly comprehensive satellite imagery on demand. The cyber war in Ukraine was also a proving ground for cyber defense firms. As a Microsoft intelligence report said, "The first shots [in the Ukraine war] were in fact fired hours before, when the calendar still said February 23. They involved a cyberweapon called 'Foxblade' that was launched against computers in Ukraine. Reflecting the technology of our time, those among the first to observe the attack were half a world away, working in the United States in Redmond, Washington."¹⁹

The fact that commercial intelligence is available to everyone is an asset as well: Sharing across borders is simpler if there is no declassification process, no calculation about revealing sources and methods. The United States does not need to protect Maxar's secrets. The easy availability of unclassified evidence probably helped prompt the Biden administration to go public with additional intelligence that indicated Russia was planning an imminent full-scale invasion. That effort galvanized Europe to overcome its own cognitive bias-a false sense of hope that Russia would leave Ukraine alone.

Israel: High-Tech Collection and Cognitive Bias

In the run-up to October 7, Israel's high-tech intelligence collection against Hamas worked; it was only the interpretation of that information that failed. At least a year before the attack, Israeli intelligence collected a copy of Hamas's attack plan, called "Jericho Wall." The plan showed how Hamas planned to take apart automated security measures, including cameras and sensors built into perimeter fences.²⁰ Months before the attack, a young female analyst wrote a report flagging that a Hamas day-long training exercise matched the stolen plan. Separately, Israel's red team unit, looking at largely the same information, issued four warnings that Hamas was planning for a confrontation.²¹ Around the same time, Egypt's intelligence service told its Israeli counterpart that "something big" was in the works.²² And the night before the attack, security services saw dozens of Israeli SIM cards activated.²³

Despite all these signs, Hamas managed to send hundreds of fighters into Israel, causing at least 1,200 casualties. The "why" of this failure will haunt Israel for decades, but early analysis boils down to a mental block, in the form of anchor bias and confirmation bias.²⁴ Humans tend to anchor their beliefs to certain information. They then use new information to "confirm" those perhaps erroneous beliefs. Unless officers work to identify and break these biases, disaster can strike even the most sophisticated intelligence service.

Israel's war on its northern border, however, was a highly effective-and lethal-combination of intelligence and warfighting. Israel pulled off a clever, tailored covert-action operation that caused more than 3,000 Hezbollah pagers to explode simultaneously, disabling the bulk of Hezbollah's fighting force and severing its communications network. Over the course of nearly 20 years, Israel had developed targeting packages against the totality of Hezbollah leadership and frontline positions. When the fight turned kinetic, the Israel Defense Forces destroyed more than 1,600 Hezbollah facilities and weapons sites. Those strikes killed four Hezbollah senior leaders, including Secretary General Hassan Nasrallah. In six weeks, Hezbollah went from the most capable terrorist group in the world to a shell of its former strength, thanks to the strength of Israeli intelligence gathering.

Both of these conflicts are instructive for an era of great power competition. China and Russia have learned lessons from the Ukraine war, including the importance of information warfare and how to capitalize on, or undermine, an engaged population serving as a network of sensors. On a larger scale, China in particular has perfected the art of staying below the threshold of antagonizing the United States while aggressively collecting its own intelligence. It also has fully committed to technological competition, pushing ahead with the next generation of technologies that will provide an immense intelligence edge.

Implications for the Future of Intelligence

The wars in Ukraine and Israel and the accelerating competition between the United States and China underscore several implications for the future of intelligence work. Information is plentiful and can be used responsibly or selectively to serve a particular viewpoint. The wars of the future will be fought in conditions of near transparency, and intelligence collection efforts will be similarly exposed to scrutiny. But just because facts are available does not mean interpreting them will be straightforward. Intelligence professionals will need to be humble about what they do not know, and they will need an extensive rolodex to find someone to help, and help quickly. The sections below explore these factors in more depth.

Dueling Facts

The oft-repeated quote "there are lies, damn lies, and statistics," popularly attributed to Mark Twain or Benjamin Disraeli, will apply in force to the modern environment featuring oceans of data. With so much available information, a person can find data to support any point of view. To use a popular example, data shows that shark attacks rise in lockstep with ice cream sales; bad data science could lead a person to assess that sharks prefer people who taste like ice cream. Intelligence assessments drawing on a seemingly endless sea of data must be rigorous in both logic and collection to avoid mistakes like mixing up correlation and causation (ice cream sales and shark attacks both go up when people spend time at the beach) or something far more serious, like deciding whether a pattern of data indicates a country is preparing for war. Decisionmakers must be discerning, work with intelligence analysts to interrogate the data, ask for confidence levels, and investigate whether contradictory evidence exists to ensure strong outcomes. Patience will also be required-solid tradecraft takes time, and the first answers are almost never the right ones. An internet sleuth could be first to the scoop-and very wrong.

AI-Enabled Insights

AI and cloud computing are empowering those in and outside government to learn more, know more, and find more. If a curious individual can ask good questions, AI can find the data and sort the results as requested. Inside intelligence services, the good questions are the easy part. The hard part is ensuring the security of the AI systems and the integrity of its answers. The even harder part is the cultural change necessary to make best use of the revolutionary technology. Fear of change is a serious friction point, and using AI as a copilot is a big change.²⁵ The U.S. intelligence community is already incorporating AI and machine learning in processing huge amounts of video and imagery. MI-6 has reportedly used AI to summarize information and sift through the ever-growing sea of data, while China's Ministry of State Security developed an AI system capable of tracking U.S. spies and other foreign agents.²⁶ The next frontier will be using AI to process and summarize quantities of text in a dependable way, with a system capable of showing its sources and protecting classified information in a high-side environment. An AI system that hallucinates preparations for a coup is exceedingly dangerous, but an AI system that can summarize 10 years of speeches in 10 minutes to analyze the decisionmaking style of a world leader is invaluable. As AI systems progress beyond data processing toward agentic decisionmaking, intelligence services will be able to send autonomous systems into hostile environments for long-dwell intelligence collection, with the system able to "decide" when it should emerge and report home.

From Toiling in the Shadows to Fighting in the Light

In the early days of the U.S. intelligence community, the National Security Agency (NSA) was referred to as "No Such Agency." The National Reconnaissance Office did not exist. Today, CIA has an account on X, formerly known as Twitter; its famous first post was rather tongue-in-cheek (see below).