Applied AI

The Secretary of War issued this strategy on January 9, 2026. It is the flagship DoW-wide AI policy document and the primary driver of current AI activity across all components. The strategy directs the Department to become an "AI-first" warfighting force by:

• Deploying commercial frontier AI models across all classification levels
• Aggressively identifying and eliminating bureaucratic barriers to data access and AI integration
• Executing seven Pace-Setting Projects (PSPs) as the new execution standard for military AI
• Focusing investment on U.S. asymmetric advantages in compute, model innovation, capital markets, and two decades of combat-proven operational data

The strategy introduces specific execution expectations: speed wins over process, AI model parity requires deploying the latest frontier models within 30 days of public release, and a monthly Barrier Removal Board holds authority to waive non-statutory requirements. Read the full strategy (PDF).

The January 2026 AI Strategy designates seven initial PSPs to establish the new execution standard for military AI. Each has a single accountable leader, aggressive timelines, and measurable outcomes. Monthly progress is reported to the Deputy Secretary of War and USD(R&E). Initial demonstrations to user partners are required within six months of the strategy's January 2026 publication date.

Warfighting

• Swarm Forge - A competitive mechanism to iteratively discover, test, and scale novel ways of fighting with and against AI-enabled capabilities, combining elite warfighting units with elite technology innovators.
• Agent Network - Unleashes AI agent development and experimentation for AI-enabled battle management and decision support, from campaign planning to kill chain execution.
• Ender's Foundry - Accelerates AI-enabled simulation capabilities and sim-dev/sim-ops feedback loops to stay ahead of AI-enabled adversaries.

Intelligence

• Open Arsenal - Accelerates the TechINT-to-capability development pipeline, turning intelligence into weapons in hours rather than years.
• Project Grant - Transforms deterrence from static postures and speculation to dynamic pressure with interpretable results.

Enterprise

• GenAI.mil - Puts commercial frontier AI models directly in the hands of all DoW civilian and military personnel across all classification levels, democratizing AI experimentation across three million users.
• Enterprise Agents - Builds the playbook for rapid and secure AI agent development and deployment to transform enterprise workflows.

The CDAO is required to make all foundational enablers unlocked by PSPs available to programs Department-wide in real time. Each Military Department, combatant command, and defense agency was directed within 30 days to identify at least three fast-follow projects aligned to the PSPs.

The CDAO is the principal DoW office responsible for accelerating the adoption of data, analytics, and AI from the enterprise to the battlefield. Under the 2026 AI Strategy, CDAO is refocused on foundational enablers: infrastructure, data, models, policies, and talent. It acts as executing authority for the Pace-Setting Projects and maintains department-wide AI metrics and deployment velocity tracking.

CDAO operates under USD(R&E), following an August 2025 realignment intended to tighten integration between AI capability development and the broader research and engineering enterprise.

Current CDAO Leadership

• Cameron Stanley - Chief Digital and Artificial Intelligence Officer
• Andrew Mapes - Acting Principal Deputy CDAO

Maven Smart System (MSS)

Maven Smart System is the DoW's primary tactical AI platform, built on Palantir's AIP software. It fuses data from multiple sensors for real-time object detection, tracking, and decision support in combat operations. MSS supports AI-enabled battle management and underpins CJADC2 efforts across the joint force.

Key developments in 2025-2026:

• In May 2024, DoW signed an initial $480 million, five-year IDIQ contract with Palantir covering five combatant commands.
• In May 2025, coverage expanded across additional commands and components, including an Army Enterprise Agreement valued at up to $10B.
• In March 2026, Deputy Secretary Feinberg directed MSS to transition to a formal program of record, with oversight consolidated at a new CDAO Maven Smart System Program Office.
• Operation Epic Fury (the 2026 air campaign against Iran) used MSS to support strike planning across 13,000 targets over 38 days, according to CDAO Cameron Stanley.
• The FY27 budget request includes $2.3B for MSS and the Joint Fires Network.
• NATO acquired MSS NATO in March 2025 for employment within Allied Command Operations.

War Data Platform (WDP)

The War Data Platform is a CDAO initiative focused on expanding the core data integration layer across DoW. WDP provides standardized data access to streamline secure, rapid development and integration of agentic AI and other applications across the Department.

Tradewinds

Tradewinds is CDAO's commercial solutions marketplace, connecting industry with DoW data and AI challenges. It serves as a primary on-ramp for companies offering AI capabilities to the Department. See Tradewinds at ai.mil.

GAMECHANGER

GAMECHANGER is an AI and natural language processing application managed by the OSD Comptroller. It enables DoW policymakers to navigate more than 38,000 policy and budget documents from 28 authoritative sources. Key capabilities include intelligent keyword search, a knowledge graph for mapping document relationships, and NLP-driven automation for identifying and integrating policy requirements. The platform has more than 7,500 users.

Department of the Air Force AI Strategy (April 2026)

Secretary of the Air Force Troy Meink signed the DAF AI Strategy in April 2026. The strategy is issued under authorities from the 2026 National Defense Strategy and the 2026 DoW AI Strategy. Its core premise: AI is not a future capability but an indispensable operational tool that must be integrated now.

Vision: The DAF is an AI-first force providing unmatched strategic advantages through capabilities that increase the speed, precision, and agility of mission threads to enhance decision-making, operational readiness, and effectiveness across the force.

Five mission areas:

• Decision Superiority in Multi-Domain Operations - Leverage AI to improve the Observe-Orient-Decide-Act loop through better observation, sense-making, and autonomous decision support in contested environments.
• Readiness and Sustainment - Optimize asset availability, predictive maintenance, supply chains, and resource management.
• Enterprise Optimization and Workforce Augmentation - Enhance back-office functions, improve personnel management, and free human capital for higher-value tasks.
• Training, Education, Modeling, and Simulation - Integrate AI into training to create realistic, adaptive, and personalized learning environments.
• Research, Development, and Modernization - Accelerate the entire defense capability lifecycle from concept to deployment.

Five strategic imperatives: Unleash the power of data; accelerate an AI-first culture; build the enterprise AI ecosystem; drive agile adoption and process reform; modernize assurance for an AI-paced world.

Five implementation building blocks: Data, technology, and infrastructure; talent and workforce; partnerships and ecosystem; change management and process re-engineering; AI governance and oversight.

The DAF CDAO leads implementation. A companion implementation plan with detailed delivery timelines was announced as forthcoming at the time of publication. Read the full DAF AI Strategy (PDF).

U.S. Space Force Data and AI FY2025 Strategic Action Plan (March 2025)

As the world's first digital service, USSF recognizes data and AI as essential to maintaining space superiority. The FY2025 Strategic Action Plan builds on the FY2024 inaugural plan and sets four Lines of Effort for the force:

• LOE 1: Mature enterprise-wide data and AI governance - Establish governance structures, field Command-level data officers, and create three working groups covering data stewardship, AI innovation and adoption, and operational testing and infrastructure data standards.
• LOE 2: Advance a data-driven and AI-enabled culture - Build AI literacy across the Guardian workforce through PME integration, professional development events, AI boot camps, and an AI innovation challenge.
• LOE 3: Rapidly adopt data, advanced analytics, and AI technologies - Transition the Unified Data Library (UDL) to a formal program of record, standardize benchmarks for LLMs in space operations contexts, and integrate critical sensor data for Space Domain Awareness analysis.
• LOE 4: Strengthen government, academic, industry, and international partnerships - Host the Space AI Symposium in partnership with the National Reconnaissance Office, advance research in compute capacity and AI-driven anomaly detection, and establish UDL API gateways for data sharing across DoD and allied partners.

Read the full USSF Data and AI FY2025 Strategic Action Plan (PDF).

The Army and Navy AI Strategies are coming soon.

Global Information Dominance Experiment (GIDE)

GIDE is a series of joint, globally integrated experiments using data, analytics, and AI to stress current systems and explore new technology for decision advantage. Originally initiated by U.S. Northern Command, GIDE is now managed by CDAO in partnership with the Joint Staff, with future events aligned to the Joint Warfighting Concept and the JADC2 Implementation Strategy.

AI Acceleration Strategy: Pace-Setting Projects

See the dedicated PSP section above. CDAO tracks and ranks the full portfolio of fast-follow AI efforts by speed and impact, reporting monthly to the Deputy Secretary and USD(R&E).

DoD Data Decrees

The January 2026 AI Strategy directs CDAO to enforce the "DoD Data Decrees" a set of data access mandates requiring all DoW components to maintain and deliver federated data catalogs exposing system interfaces, data assets, and access mechanisms across all classification levels. Denials of CDAO data requests must be justified to USD(R&E) within seven days. This is intended to eliminate data hoarding as a structural barrier to AI exploitation.

Artificial Intelligence (AI) - A broad term for software systems that perform tasks traditionally requiring human intelligence: perceiving, reasoning, learning, problem solving, and decision-making. DoW strategy defines the goal as fielding AI that enables machines to perform tasks more quickly and accurately than human operators in specific mission contexts.

Machine Learning (ML) - A subfield of AI in which systems learn from data rather than explicit programming. A developer selects an algorithm, provides training data, and the system learns to make predictions or decisions. ML is the foundation of most current military AI applications, from target recognition to logistics optimization.

Deep Learning - A subset of ML using multi-layer neural networks to process complex data (images, audio, text). Deep learning achieves high performance on tasks like object detection in satellite imagery, speech recognition, and natural language processing.

Generative AI - AI models that produce content (text, code, images) in response to a prompt. Large language models like those deployed through GenAI.mil and Maven Smart System are generative AI applications. DoW strategy directs deployment of frontier generative AI models across all classification levels.

Narrow AI - Systems that perform only the specific task they were trained for. All currently deployed military AI is narrow AI. Artificial general intelligence (AGI) does not yet exist.

Autonomy - The engineering discipline expanding robots' and vehicles' ability to perform tasks with limited human interaction. Distinct from AI but closely related: autonomous systems depend on AI and ML for perception and decision-making.

AI Failure Modes

Program managers, testers, and operators need to understand that ML systems have failure modes distinct from traditional software. Key risks include:

• Algorithmic bias - Systems trained on unrepresentative data can produce discriminatory or operationally incorrect outputs. Facial recognition programs have exhibited racial bias due to insufficient training data diversity.
• Adversarial vulnerabilities - Subtle, often imperceptible input modifications can cause AI systems to misclassify targets with high confidence. This creates exploitable vulnerabilities in AI-assisted targeting and identification.
• Distribution shift - A system trained in one environment may fail when deployed in different conditions (weather, terrain, adversary tactics) it was not trained to handle.
• Generative AI hallucination - Large language models can produce confidently stated but factually incorrect outputs. Human review remains essential for any decision with significant consequences.

ML Algorithm Families

• Supervised Learning - Uses labeled training data. High performance but requires significant labeled data investment.
• Unsupervised Learning - Uses unlabeled data. Lower performance for many tasks but applicable where labeled data is unavailable.
• Semi-Supervised Learning - Combines labeled and unlabeled data.
• Reinforcement Learning - AI agents learn through trial-and-error interaction with an environment. Significant research potential; harder to deploy in complex real-world environments.

Military AI applications span the full mission spectrum. Current and near-term applications include:

• Intelligence, Surveillance, and Reconnaissance (ISR) - Automated target detection in optical and SAR imagery; anomaly detection in radio frequency signatures; fusing imaging and chemical or biological sensors for threat detection. ML consistently outperforms human analysts for finding targets in high-clutter environments.
• Command and Control - AI-enabled battle management from campaign planning to kill chain execution (the Agent Network PSP). Decision support tools that compress sensor-to-shooter timelines (the core CJADC2 application of Maven Smart System).
• Autonomous Systems - Drone swarms and ground vehicles that coordinate attacks, ISR missions, and logistics operations through collaborative autonomy. The Swarm Forge PSP is specifically focused on discovering and scaling these concepts.
• Electronic Warfare and Cyber - ML countermeasures against low-probability-of-intercept radar; behavioral modeling of unknown radar emitters to infer intent and predict future actions.
• Space Domain Awareness - Predicting spacecraft maneuvers, characterizing possible actions, and assessing threat trajectories. USSF's Unified Data Library and Space Domain Awareness programs depend on AI for near-real-time analysis.
• Logistics and Readiness - Predictive maintenance to prevent equipment failures before they occur; logistics optimization to reduce costs and improve readiness rates. The DAF AI Strategy specifically identifies these as near-term AI imperatives.
• Simulation and Training - AI-enabled simulation (Ender's Foundry PSP) to generate realistic, adaptive training environments and accelerate tactics development without live-fire cost or risk.

Source: CRS Emerging Technologies Jan 2026

Although the U.S. government has no official definition of artificial intelligence, policymakers generally use the term AI to refer to a computer system capable of human-level cognition. AI is further divided into three categories: narrow AI, artificial general intelligence (AGI), and artificial superintelligence. Narrow AI systems can perform only the specific task that they were trained to perform, while AGI systems would be capable of performing a broad range of tasks, including those for which they were not specifically trained. Artificial superintelligence refers to a system that could exceed human-level cognition across most tasks. AGI systems and artificial superintelligence do not yet—and may never—exist.

Narrow AI is currently being incorporated into a number of military applications by both the United States and its competitors. Such applications include but are not limited to intelligence, surveillance, and reconnaissance; logistics; cyber operations; command and control; and semi-autonomous and autonomous vehicles. These technologies are intended in part to augment or replace human operators, freeing them to perform more complex and cognitively demanding work. In addition, AI-enabled systems could (1) react significantly faster than systems that rely on operator input, (2) cope with an exponential increase in the amount of data available for analysis, and (3) enable new concepts of operations, such as swarming (i.e., cooperative behavior in which uncrewed vehicles autonomously coordinate to achieve a task) that could confer a warfighting advantage by overwhelming adversary defensive systems.

Narrow AI could, however, introduce a number of challenges. For example, such systems may be subject to algorithmic bias as a result of their training data or models. Researchers have repeatedly discovered instances of racial bias in AI facial recognition programs due to the lack of diversity in the images on which the systems were trained, while some natural language processing programs have developed gender bias. Such biases could hold significant implications for AI applications in a military context. A number of U.S. government documents, including the Pentagon's Responsible Artificial Intelligence Strategy and Implementation Pathway, provide guidance on these applications.

Source: Emerging Military Technology, CRS, Oct 2021

Although the U.S. government has no official definition of artificial intelligence, policymakers generally use the term AI to refer to a computer system capable of human-level cognition. AI is further divided into two categories: narrow AI and general AI.

• Narrow AI systems can perform only the specific task that they were trained to perform,
• General AI systems would be capable of performing a broad range of tasks, including those for which they were not specifically trained. General AI systems do not yet—and may never—exist.

Narrow AI is currently being incorporated into a number of military applications by both the United States and its competitors. Such applications include but are not limited to intelligence, surveillance, and reconnaissance logistics; cyber operations; command and control; and semiautonomous and autonomous vehicles. These technologies are intended in part to augment or replace human operators, freeing them to perform more complex and cognitively demanding work. In addition, AI-enabled systems could

1. react significantly faster than systems that rely on operator input;
2. cope with an exponential increase in the amount of data available for analysis
3. enable new concepts of operations, such as swarming (i.e., cooperative behavior in which unmanned vehicles autonomously coordinate to achieve a task) that could confer a warfighting advantage by overwhelming adversary defensive systems.

Narrow AI, however, could introduce a number of challenges. For example, such systems may be subject to algorithmic bias as a result of their training data or models. Researchers have repeatedly discovered instances of racial bias in AI facial recognition programs due to the lack of diversity in the images on which the systems were trained, while some natural language processing programs have developed gender bias. Such biases could hold significant implications for AI applications in a military context. For example, incorporating undetected biases into systems with lethal effects could lead to cases of mistaken identity and the unintended killing of civilians or noncombatants.

Similarly, narrow AI algorithms can produce unpredictable and unconventional results that could lead to unexpected failures if incorporated into military systems. In a commonly cited demonstration of this phenomenon, researchers combined a picture that an AI system correctly identified as a panda with random distortion that the computer labeled “nematode.” The difference in the combined image is imperceptible to the human eye, but it resulted in the AI system labeling the image as a gibbon with 99.3% confidence. Such vulnerabilities could be exploited intentionally by adversaries to disrupt AI-reliant or -assisted target identification, selection, and  engagement. This could, in turn, raise ethical concerns—or, potentially, lead to violations of the law of armed conflict—if it results in the system selecting and engaging a target or class of targets that was not approved by a human operator.

Finally, recent news reports and analyses have highlighted the role of AI in enabling increasingly realistic photo, audio, and video digital forgeries, popularly known as “deep fakes.” Adversaries could deploy this AI capability as part of their information operations in a “gray zone” conflict. Deep fake technology could be used against the United States and its allies to generate false news reports, influence public discourse, erode public trust, and attempt blackmail of government officials. For this reason, some analysts argue that social media platforms—in addition to deploying deep fake detection tools—may need to expand the means of labeling and authenticating content. Doing so might require that users identify the time and location at which the content originated or properly label content that has been edited. Other analysts have expressed concern that regulating deep fake technology could impose an undue burden on social media platforms or lead to unconstitutional restrictions on free speech and artistic expression. These analysts have suggested that existing law is sufficient for managing the malicious use of deep fakes and that the focus should be instead on the need to educate the public about deep fakes and minimize incentives for creators of malicious deep fakes.

Source: NSCAI Final Report, Mar 2021

The National Security Commission on Artificial Intelligence (NSCAI) humbly acknowledges how much remains to be discovered about AI and its future applications. Nevertheless, we know enough about AI today to begin with two convictions. First, the rapidly improving ability of computer systems to solve problems and to perform tasks that would otherwise require human intelligence—and in some instances exceed human performance—is world altering. AI technologies are the most powerful tools in generations for expanding knowledge, increasing prosperity, and enriching the human experience. AI is also the quintessential “dual-use” technology. The ability of a machine to perceive, evaluate, and act more quickly and accurately than a human represents a competitive advantage in any field—civilian or military. AI technologies will be a source of enormous power for the companies and countries that harness them.

Second, AI is expanding the window of vulnerability the United States has already entered. For the first time since World War II, America’s technological predominance—the backbone of its economic and military power—is under threat. China possesses the might, talent, and ambition to surpass the United States as the world’s leader in AI in the next decade if current trends do not change. Simultaneously, AI is deepening the threat posed by cyber attacks and disinformation campaigns that Russia, China, and others are using to infiltrate our society, steal our data, and interfere in our democracy. The limited uses of AI-enabled attacks to date represent the tip of the iceberg. Meanwhile, global crises exemplified by the COVID-19 pandemic and climate change highlight the need to expand our conception of national security and find innovative AI-enabled solutions.

Part I: Defending America in the AI Era.

• Defend against emerging AI-enabled threats to America’s free and open society.
• Prepare for future warfare
• Manage risks associated with AI-enabled and autonomous weapons
• Transform national intelligence.
• Scale up digital talent in government.
• Establish justified confidence in AI systems.
• Present a democratic model of AI use for national security

Part II: Winning the Technology Competition.

• Organize with a White House–led strategy for technology competition
• Win the global talent competition
• Accelerate AI innovation at home
• Implement comprehensive intellectual property (IP) policies and regimes
• Build a resilient domestic base for designing and fabricating microelectronics.
• Protect America’s technology advantages
• Build a favorable international technology order
• Win the associated technologies competitions.

Conclusion

This new era of competition promises to change the world we live in and how we live within it. We can either shape the change to come or be swept along by it. We now know that the uses of AI in all aspects of life will grow and the pace of innovation will continue to accelerate. We know adversaries are determined to turn AI capabilities against us. We know China is determined to surpass us in AI leadership. We know advances in AI build on themselves and confer significant first mover advantages. Now we must act. The principles we establish, the federal investments we make, the national security applications we field, the organizations we redesign, the partnerships we forge, the coalitions we build, and the talent we cultivate will set America’s strategic course. The United States should invest what it takes to maintain its innovation leadership, to responsibly use AI to defend free people and free societies, and to advance the frontiers of science for the benefit of all humanity. AI is going to reorganize the world. America must lead the charge.

Read the full NSCAI Final Report

Expert knowledge.

The first and oldest type of AI whereby a computer is programmed with detailed rules based on human expertise or criteria and produces outputs consistent with its programming. An example of such rules-based AI capabilities for DOD is maintenance software for aircraft that requires users to input their information according to prespecified data formats and then processes that data according to rules programmed by human experts (i.e., maintenance professionals) to diagnose the cause of malfunctions.

Machine learning.

The second and current type of AI whereby a computer is given basic instructions and fed training data to learn how to predict specific outcomes. According to an academic publication, machine learning AI is an appropriate solution when writing a program for a machine to follow is too time-consuming or otherwise not possible.9 Instead of explicit programming, this type of AI requires a developer to select an appropriate algorithm based on the desired result, feed it the appropriate training data, and watch to see if the algorithm learns what it is supposed to. If the AI model is not performing as expected, the developer can revise the training data, adjust the algorithm parameters, or chose a different algorithm. An example for DOD is facial recognition technology that uses a set of algorithms to identify individuals by instantaneously searching databases of faces and comparing them to those detected in a video or photograph.

Contextual adaptation.

The third and potential future type of AI whereby a computer is capable of adapting to new situations without needing to be retrained while also being able to explain to users the reasoning behind its decisions or predictions. A potential example for DOD is a fully autonomous ship that uses algorithms to maneuver in situations it was not specifically trained for (such as inclement weather or contested waters) and is capable of planning, relaying, and carrying out military missions similar to the way a human would.

Source: GAO-22-104765 Status of Developing and Acquiring AI Capabilities for Weapon Systems