U.S. Diplomacy
Since its starting in 1922, Foreign Affairs has been the leading forum for serious discussion of American diplomacy and international affairs. The publication has included contributions from numerous leading global affairs specialists.

More Resources

- Feedback
- Institutional Subscriptions
- Gift a Membership
- 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 genbecle.com Cyber and Emerging Technology on the U.S. National Security Council. From 2009 to 2021, she served in senior functional functions in intelligence and cybersecurity at the National Security Agency, including as its very 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 feature. Subscribe now or Sign in.


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


Download PDF. This is a subscriber-only function. Subscribe now or Check in.


Request Reprint. Request reprint approvals here.




In the early 1950s, the United States dealt with a vital intelligence challenge in its growing competition with the Soviet Union. Outdated German reconnaissance images from World War II could no longer provide adequate intelligence about Soviet military abilities, and existing U.S. security abilities were no longer able to permeate the Soviet Union’s closed airspace. This shortage spurred an audacious moonshot effort: the advancement of the U-2 reconnaissance aircraft. In just a few years, U-2 objectives were delivering essential intelligence, recording images of Soviet missile 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 juncture. Competition between Washington and its competitors over the future of the global order is magnifying, and now, much as in the early 1950s, the United States should make the most of its world-class private sector and ample capacity for innovation to outcompete its adversaries. The U.S. intelligence community need to harness the country’s sources of strength to deliver insights to policymakers at the speed these days’s world. The combination of expert system, especially through big language designs, offers groundbreaking chances to improve intelligence operations and analysis, making it possible for the shipment of faster and more pertinent support to decisionmakers. This technological transformation features substantial downsides, however, especially as adversaries make use of comparable developments to discover and counter U.S. intelligence operations. With an AI race underway, the United States need to challenge itself to be first-first to gain from AI, first to secure itself from enemies who might utilize the innovation for ill, and initially to use AI in line with the laws and worths of a democracy.

For the U.S. national security community, satisfying the guarantee and handling the peril of AI will need deep technological and cultural modifications and a determination to alter the method companies work. The U.S. intelligence and military neighborhoods can harness the capacity of AI while alleviating its intrinsic threats, guaranteeing that the United States maintains its one-upmanship in a rapidly progressing global landscape. Even as it does so, the United States need to transparently communicate to the American public, and to populations and partners all over the world, how the country intends to fairly and securely use AI, in compliance with its laws and worths.

MORE, BETTER, FASTER

AI‘s potential to transform the intelligence neighborhood depends on its ability to process and evaluate large amounts of information at unprecedented speeds. It can be challenging to analyze large quantities of gathered information to create time-sensitive warnings. U.S. intelligence services might utilize AI systems’ pattern recognition capabilities to determine and alert human analysts to prospective threats, such as rocket launches or military movements, or crucial global developments that analysts understand senior U.S. decisionmakers have an interest in. This capability would guarantee that critical cautions are prompt, actionable, and pertinent, permitting more effective reactions to both rapidly emerging hazards and emerging policy chances. Multimodal designs, which integrate text, images, and audio, improve this analysis. For example, using AI to cross-reference satellite imagery with signals intelligence might offer a detailed view of military movements, making it possible for faster and more accurate risk assessments and potentially brand-new means of delivering details to policymakers.

Intelligence analysts can likewise offload recurring and lengthy jobs to makers to concentrate on the most satisfying work: generating initial and much deeper analysis, increasing the intelligence neighborhood’s total insights and efficiency. An excellent example of this is foreign language translation. U.S. intelligence companies invested early in AI-powered abilities, and the bet has paid off. The capabilities of language designs have actually grown progressively sophisticated and accurate-OpenAI’s just recently released o1 and o3 models showed substantial development in accuracy and reasoning ability-and can be utilized to even more quickly equate and summarize text, audio, wolvesbaneuo.com and video files.

Although challenges remain, future systems trained on greater amounts of non-English data could be capable of critical subtle differences in between dialects and comprehending the meaning and cultural context of slang or Internet memes. By depending on these tools, the intelligence neighborhood might focus on training a cadre of extremely specialized linguists, who can be hard to find, typically battle to make it through the clearance procedure, and take a long period of time to train. And naturally, by making more foreign language products available throughout the right agencies, U.S. intelligence services would have the ability to more rapidly triage the mountain of foreign intelligence they receive to choose the needles in the haystack that truly matter.

The value of such speed to policymakers can not be underestimated. Models can quickly sort through intelligence data sets, open-source details, and conventional human intelligence and produce draft summaries or initial analytical reports that experts can then validate and improve, making sure the end products are both detailed and precise. Analysts might coordinate with a sophisticated AI assistant to overcome analytical issues, test ideas, and brainstorm in a collective fashion, improving each model of their analyses and delivering finished intelligence more rapidly.

Consider Israel’s experience in January 2018, when its intelligence service, the Mossad, covertly burglarized a secret Iranian facility and stole 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 an additional 55,000 files saved on CDs, consisting of images and videos-nearly all in Farsi. Once the archive was obtained, senior authorities put immense pressure on intelligence experts to produce detailed assessments of its material and whether it indicated an ongoing effort to develop an Iranian bomb. But it took these professionals a number of months-and numerous hours of labor-to equate each page, review it by hand for appropriate material, and include that details into evaluations. With today’s AI capabilities, the first 2 actions in that procedure could have been accomplished within days, perhaps even hours, enabling analysts to comprehend and contextualize the intelligence rapidly.

Among the most interesting applications is the way AI could transform how intelligence is taken in by policymakers, enabling them to communicate straight with intelligence reports through ChatGPT-like platforms. Such abilities would allow users to ask particular concerns and receive summarized, appropriate details from thousands of reports with source citations, assisting them make informed decisions rapidly.

BRAVE NEW WORLD

Although AI provides various advantages, it likewise poses significant new threats, especially as enemies establish similar innovations. China’s developments in AI, particularly in computer vision and security, threaten U.S. intelligence operations. Because the country is ruled by an authoritarian routine, it does not have privacy constraints and civil liberty protections. That deficit makes it possible for massive information collection practices that have yielded data sets of immense size. Government-sanctioned AI models are trained on vast amounts of individual and behavioral information that can then be used for numerous functions, such as security and social control. The existence of Chinese business, such as Huawei, in telecoms systems and software application around the globe could provide China with prepared access to bulk information, notably bulk images that can be used to train facial acknowledgment designs, a particular concern in nations with big U.S. military bases. The U.S. nationwide security neighborhood need to consider how Chinese models constructed on such comprehensive information sets can provide China a strategic advantage.

And it is not just 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 company DeepSeek, is putting effective AI capabilities into the hands of users around the world at fairly economical costs. A lot of these users are benign, however some are not-including authoritarian regimes, cyber-hackers, and criminal gangs. These malign actors are using large language models to rapidly produce and spread out incorrect and malicious material or to conduct cyberattacks. As experienced with other intelligence-related innovations, such as signals obstruct capabilities and unmanned drones, China, Iran, and Russia will have every incentive to share some of their AI developments with client states and subnational groups, such as Hezbollah, Hamas, and the Wagner paramilitary business, therefore increasing the danger to the United States and its allies.

The U.S. military and intelligence neighborhood’s AI designs will end up being appealing targets for foes. As they grow more effective and main to U.S. nationwide security decision-making, intelligence AIs will become crucial nationwide properties that should be defended against foes looking for to compromise or control them. The intelligence community should buy establishing safe AI models and in developing requirements for “red teaming” and continuous assessment to secure against prospective risks. These teams can utilize AI to simulate attacks, revealing possible weaknesses and establishing methods to reduce them. Proactive procedures, including cooperation with allies on and financial investment in counter-AI technologies, will be vital.

THE NEW NORMAL

These obstacles can not be wished away. Waiting too long for AI technologies to completely mature brings its own threats