Spy Vs. AI

U.S. Foreign Policy


Since its founding in 1922, Foreign Affairs has actually been the leading forum for serious conversation of American diplomacy and worldwide affairs. The magazine has featured contributions from numerous prominent global affairs experts.


More Resources


- Feedback

- Institutional Subscriptions

- Gift a Subscription

- About Us

- Events

- Issue Archive

- Advertise

- Audio Content

- Account Management

- FAQs


Spy vs. AI


ANNE NEUBERGER is Deputy Assistant to the President and Deputy National Security Adviser for Cyber and Emerging Technology on the U.S. National Security Council. From 2009 to 2021, she served in senior operational functions in intelligence and cybersecurity at the National Security Agency, consisting of as its first Chief Risk Officer.


- More by Anne Neuberger


Spy vs. AI


How Artificial Intelligence Will Remake Espionage


Anne Neuberger


-.

Copy Link Copied.
Article link: https://www.foreignaffairs.com/united-states/spy-vs-aihttps://www.foreignaffairs.com/united-states/spy-vs-ai.
Copy



Gift Link Copied.
This is a subscriber-only function. Subscribe now or Check in.



Create Citation Copied.
Chicago MLA APSA APA.
Chicago Cite not available at the minute.
MLA Cite not available at the moment.
APSA Cite not available at the minute.
APA Cite not available at the minute



Download PDF.
This is a subscriber-only feature. Subscribe now or Sign in.



Request Reprint.
Request reprint permissions here.






In the early 1950s, the United States faced a critical intelligence challenge in its burgeoning competition with the Soviet Union. Outdated German reconnaissance photos from World War II might no longer provide sufficient intelligence about Soviet military capabilities, and existing U.S. security abilities were no longer able to penetrate the Soviet Union's closed airspace. This shortage stimulated an audacious moonshot initiative: the advancement of the U-2 reconnaissance aircraft. In just a couple of years, U-2 missions were delivering important intelligence, surgiteams.com recording pictures of Soviet rocket installations in Cuba and bringing near-real-time insights from behind the Iron Curtain to the Oval Office.


Today, the United States stands at a similar point. Competition in between Washington and its rivals over the future of the worldwide order is magnifying, and now, much as in the early 1950s, the United States need to make the most of its first-rate private sector and adequate capacity for innovation to outcompete its adversaries. The U.S. intelligence community must harness the nation's sources of strength to provide insights to policymakers at the speed these days's world. The combination of expert system, particularly through big language models, uses groundbreaking opportunities to enhance intelligence operations and analysis, making it possible for the shipment of faster and more appropriate assistance to decisionmakers. This technological revolution includes considerable disadvantages, nevertheless, specifically as enemies exploit comparable improvements to uncover and counter U.S. intelligence operations. With an AI race underway, the United States must challenge itself to be first-first to gain from AI, first to safeguard itself from opponents who may use the technology for ill, and first to use AI in line with the laws and worths of a democracy.


For the U.S. nationwide security community, satisfying the promise and handling the hazard of AI will need deep technological and cultural modifications and a desire to change the way agencies work. The U.S. intelligence and military neighborhoods can harness the potential of AI while reducing its intrinsic threats, ensuring that the United States maintains its one-upmanship in a quickly progressing international landscape. Even as it does so, the United States must transparently convey to the American public, and to populations and partners all over the world, how the nation means to fairly and securely use AI, in compliance with its laws and worths.


MORE, BETTER, FASTER


AI's potential to reinvent the intelligence community lies in its ability to procedure and analyze vast quantities of information at unmatched speeds. It can be challenging to analyze large quantities of collected data to create time-sensitive warnings. U.S. intelligence services might leverage AI systems' pattern recognition capabilities to identify and alert human analysts to potential dangers, such as rocket launches or military motions, or crucial global advancements that experts understand senior U.S. decisionmakers have an interest in. This ability would guarantee that critical warnings are timely, actionable, and pertinent, allowing for more efficient reactions to both quickly emerging hazards and emerging policy opportunities. Multimodal designs, which integrate text, images, and audio, enhance this analysis. For circumstances, using AI to cross-reference satellite images with signals intelligence might supply a detailed view of military movements, enabling much faster and more precise hazard evaluations and possibly brand-new methods of providing details to policymakers.


Intelligence experts can also offload repetitive and lengthy jobs to machines to focus on the most fulfilling work: creating initial and much deeper analysis, increasing the intelligence community's general insights and productivity. An excellent example of this is foreign language translation. U.S. intelligence agencies invested early in AI-powered abilities, and the bet has paid off. The capabilities of language models have actually grown increasingly sophisticated and accurate-OpenAI's recently launched o1 and o3 models demonstrated substantial progress in accuracy and thinking ability-and can be utilized to much more quickly equate and sum up text, audio, and video files.


Although obstacles remain, future systems trained on higher quantities of non-English information could be capable of discerning subtle differences between dialects and understanding the meaning and cultural context of slang or Internet memes. By depending on these tools, the intelligence community might concentrate on training a cadre of extremely specialized linguists, who can be difficult to discover, frequently struggle to survive the clearance procedure, and take a long time to train. And obviously, by making more foreign language materials available across the right companies, U.S. intelligence services would be able to faster triage the mountain of foreign intelligence they receive to select the needles in the haystack that truly matter.


The worth of such speed to policymakers can not be underestimated. Models can swiftly sort through intelligence information sets, open-source details, and conventional human intelligence and produce draft summaries or preliminary analytical reports that analysts can then confirm and refine, guaranteeing the end products are both detailed and precise. Analysts could partner with a sophisticated AI assistant to resolve analytical problems, test concepts, and brainstorm in a collaborative fashion, improving each model of their analyses and providing completed intelligence more rapidly.


Consider Israel's experience in January 2018, when its intelligence service, the Mossad, discreetly broke into a secret Iranian facility and took about 20 percent of the archives that detailed Iran's nuclear activities between 1999 and 2003. According to Israeli authorities, the Mossad gathered some 55,000 pages of documents and a more 55,000 files kept on CDs, consisting of pictures and videos-nearly all in Farsi. Once the archive was obtained, senior officials positioned immense pressure on intelligence professionals to produce detailed assessments of its content and whether it pointed to an ongoing effort to build an Iranian bomb. But it took these professionals several months-and hundreds of hours of labor-to equate each page, evaluate it by hand for relevant material, and incorporate that details into evaluations. With today's AI capabilities, the very first two actions in that procedure might have been achieved within days, maybe even hours, enabling analysts to understand and contextualize the intelligence rapidly.


Among the most fascinating applications is the method AI might change how intelligence is taken in by policymakers, enabling them to connect straight with intelligence reports through ChatGPT-like platforms. Such capabilities would allow users to ask specific concerns and get summarized, pertinent details from countless reports with source citations, assisting them make informed choices quickly.


BRAVE NEW WORLD


Although AI uses many benefits, it also poses significant new threats, especially as adversaries develop comparable innovations. China's advancements in AI, especially in computer vision and security, threaten U.S. intelligence operations. Because the country is ruled by an authoritarian regime, it lacks privacy constraints and civil liberty defenses. That deficit enables large-scale information collection practices that have yielded data sets of tremendous size. Government-sanctioned AI designs are trained on large amounts of personal and behavioral data that can then be utilized for various purposes, such as security and social control. The presence of Chinese companies, such as Huawei, in telecoms systems and software all over the world might supply China with ready access to bulk data, notably bulk images that can be utilized to train facial recognition designs, a particular issue in nations with big U.S. military bases. The U.S. nationwide security neighborhood should consider how Chinese designs built on such comprehensive data sets can offer China a tactical advantage.


And it is not simply China. The proliferation of "open source" AI designs, such as Meta's Llama and those developed by the French company Mistral AI and the Chinese business DeepSeek, is putting effective AI abilities into the hands of users around the world at fairly budget friendly costs. Many of these users are benign, but some are not-including authoritarian regimes, cyber-hackers, and criminal gangs. These malign actors are utilizing big language designs to rapidly produce and spread out false and harmful material or to perform cyberattacks. As experienced with other intelligence-related innovations, such as signals intercept abilities and unmanned drones, China, Iran, and Russia will have every reward to share some of their AI developments with client states and subnational groups, such as Hezbollah, Hamas, and the Wagner paramilitary business, thus increasing the risk to the United States and its allies.


The U.S. military and intelligence community's AI designs will end up being attractive targets for foes. As they grow more powerful and main to U.S. national security decision-making, intelligence AIs will become crucial nationwide properties that need to be safeguarded against enemies seeking to jeopardize or control them. The intelligence community should buy establishing safe and secure AI models and in developing standards for "red teaming" and continuous evaluation to against possible risks. These groups can utilize AI to replicate attacks, discovering prospective weak points and establishing techniques to reduce them. Proactive measures, including collaboration with allies on and financial investment in counter-AI technologies, will be important.


THE NEW NORMAL


These difficulties can not be wanted away. Waiting too long for AI technologies to fully mature brings its own dangers; U.S. intelligence capabilities will fall back those of China, Russia, and other powers that are going complete steam ahead in developing AI. To guarantee that intelligence-whether time-sensitive warnings or longer-term strategic insight-continues to be a benefit for the United States and its allies, the nation's intelligence neighborhood needs to adjust and innovate. The intelligence services should rapidly master using AI innovations and make AI a foundational component in their work. This is the only sure way to guarantee that future U.S. presidents get the best possible intelligence support, remain ahead of their enemies, and safeguard the United States' delicate abilities and operations. Implementing these changes will need a cultural shift within the intelligence neighborhood. Today, intelligence analysts mainly develop items from raw intelligence and information, with some support from existing AI designs for voice and images analysis. Moving on, intelligence authorities should explore including a hybrid method, in line with existing laws, using AI designs trained on unclassified commercially available information and refined with classified details. This amalgam of innovation and conventional intelligence event could lead to an AI entity offering instructions to imagery, signals, open source, and measurement systems on the basis of an incorporated view of typical and anomalous activity, automated images analysis, and automatic voice translation.


To speed up the shift, intelligence leaders should promote the advantages of AI integration, highlighting the enhanced abilities and performance it provides. The cadre of newly appointed chief AI officers has been developed in U.S. intelligence and defense to work as leads within their agencies for promoting AI development and removing barriers to the technology's execution. Pilot jobs and early wins can develop momentum and self-confidence in AI's abilities, motivating wider adoption. These officers can utilize the competence of national laboratories and other partners to test and refine AI models, ensuring their efficiency and security. To institutionalize modification, leaders ought to produce other organizational incentives, including promos and training opportunities, to reward innovative methods and those workers and units that demonstrate effective use of AI.


The White House has actually developed the policy required for making use of AI in nationwide security companies. President Joe Biden's 2023 executive order regarding safe, protected, and credible AI detailed the guidance required to fairly and securely utilize the innovation, and National Security Memorandum 25, issued in October 2024, is the country's foundational strategy for utilizing the power and managing the risks of AI to advance nationwide security. Now, Congress will require to do its part. Appropriations are required for departments and companies to create the facilities needed for innovation and experimentation, conduct and scale pilot activities and assessments, and continue to buy evaluation abilities to make sure that the United States is constructing reputable and high-performing AI technologies.


Intelligence and military neighborhoods are devoted to keeping humans at the heart of AI-assisted decision-making and have actually created the frameworks and tools to do so. Agencies will need standards for how their analysts ought to utilize AI designs to make certain that intelligence items fulfill the intelligence neighborhood's requirements for dependability. The government will also need to maintain clear assistance for managing the information of U.S. citizens when it pertains to the training and usage of large language models. It will be necessary to stabilize the use of emerging innovations with safeguarding the personal privacy and civil liberties of citizens. This indicates enhancing oversight mechanisms, upgrading appropriate structures to show the abilities and risks of AI, and cultivating a culture of AI advancement within the national security device that utilizes the capacity of the technology while safeguarding the rights and flexibilities that are fundamental to American society.


Unlike the 1950s, archmageriseswiki.com when U.S. intelligence raced to the leading edge of overhead and satellite imagery by developing many of the crucial technologies itself, winning the AI race will need that neighborhood to reimagine how it partners with personal market. The economic sector, which is the main ways through which the federal government can recognize AI progress at scale, is investing billions of dollars in AI-related research, information centers, and calculating power. Given those companies' advancements, intelligence agencies need to focus on leveraging commercially available AI models and fine-tuning them with categorized information. This method makes it possible for the intelligence neighborhood to quickly expand its capabilities without having to go back to square one, allowing it to remain competitive with adversaries. A current cooperation between NASA and IBM to develop the world's biggest geospatial foundation model-and the subsequent release of the model to the AI community as an open-source project-is an exemplary presentation of how this kind of public-private collaboration can work in practice.


As the nationwide security neighborhood incorporates AI into its work, it should guarantee the security and durability of its designs. Establishing standards to deploy generative AI safely is essential for maintaining the stability of AI-driven intelligence operations. This is a core focus of the National Security Agency's new AI Security Center and its cooperation with the Department of Commerce's AI Safety Institute.


As the United States deals with growing rivalry to form the future of the worldwide order, it is immediate that its intelligence agencies and military profit from the country's innovation and leadership in AI, focusing particularly on big language designs, to supply faster and more pertinent details to policymakers. Only then will they gain the speed, breadth, and depth of insight needed to browse a more complicated, competitive, and content-rich world.