Project Maven
AI military intelligence program
From Wikipedia, the free encyclopedia
Project Maven (officially Algorithmic Warfare Cross Functional Team) is a Department of Defense initiative launched in April 2017 to accelerate the adoption of machine learning and data integration across U.S. military intelligence workflows[clarify], initially focused on applying computer vision for processing images and videos for intelligence purposes. Currently, the program operates under the National Geospatial-Intelligence Agency (NGA) and encompasses multiple applications across the Department of Defense spanning military operation targeting support, data integration and visualization for analysts, and training machine learning models on labeled datasets of military assets and infrastructure. It integrates data from drones, satellites, and other sensors to flag potential targets, present findings to human analysts, and relay their decisions to operational systems.
| Abbreviation | AWCFT |
|---|---|
| Nickname | Project Maven |
| Formation | April 26, 2017 |
| Founder | Robert O. Work |
| Type | Department of Defense cross-functional team |
| Legal status | Active |
| Purpose | Adoption of machine learning and data integration for military intelligence workflows |
Key people | Col. Drew Cukor Lt. Gen. Jack Shanahan |
Parent organization | National Geospatial-Intelligence Agency |
The program originated under Deputy Secretary Robert O. Work after he raised concerns about China's advances in defense applications of artificial intelligence. Project leaders, Col. Drew Cukor and Lt. Gen. Jack Shanahan, framed the program as human-in-the-loop decision support inside the Department of Defense rather than as an autonomous weapons platform.
Contractors supporting Maven have included Google, which withdrew in 2018 after internal protests, and follow-on integrators such as Palantir, Anduril, Amazon Web Services, Anthropic, and more.
The Pentagon credits Maven with providing 2024 targeting support for U.S. airstrikes in Iraq, Syria, and Yemen, along with locating hostile maritime assets in the Red Sea.
Administrative history
Initially, the effort was led by Robert O. Work who was concerned about China's military use of the emerging technology.[1] Reportedly, Pentagon development stops short of acting as an AI weapons system capable of firing on self-designated targets.[2] The project was established in a memo by the U.S. Deputy Secretary of Defense on 26 April 2017 proposing an "Algorithmic Warfare Cross-Functional Team".[3] With the help of Defense Innovation Unit, the project obtained the support of top talents in AI outside of the traditional defense contracting base. It was initially funded for $70 million.[4]
At the second Defense One Tech Summit in July 2017, Cukor also said that the investment in a "deliberate workflow process" was funded by the Department [of Defense] through its "rapid acquisition authorities" for about "the next 36 months".[5]
According to Lt. Gen. of the United States Air Force Jack Shanahan in November 2017, it is "designed to be that pilot project, that pathfinder, that spark that kindles the flame front of artificial intelligence across the rest of the [Defense] Department".[4] Its chief, U.S. Marine Corps Col. Drew Cukor, said: "People and computers will work symbiotically to increase the ability of weapon systems to detect objects."[6] Project Maven has been noted by allies, such as Australia's Ian Langford, for the ability to identify adversaries by harvesting data from sensors on UAVs and satellite.[7]
As of 2017 December, 150,000 images had been manually labelled in to establish the first training data sets, and it was projected to reach 1 million by January 2018.[4]
At the GEOINT Symposium of 2022, it was announced that Project Maven was transferred from the Office of the Under Secretary of Defense for Intelligence and Security to the NGA, under President Biden’s proposed budget for Fiscal Year 2023.[8][9] It became a Program of Record on 2023 November 7.[10]
As of 2024, the project is jointly administered by the NGA and the CDAO, and its director is Rachel Martin.[11][12] Before 2025, Biden appointees within CDAO had held back AI development for safety and reliability concerns, though as of 2025, this has stopped.[13] As of 2024, Maven provided the cloud infrastructure, software capabilities, and AI for CDAO's Combined Joint All-Domain Command and Control initiatives.[14]
On 25 March 2025, NATO Communications and Information Agency and Palantir finalized the acquisition of the Palantir Maven Smart System NATO (MSS NATO) for employment within NATO's Allied Command Operations. It was planned to be used within 30 days of acquisition.[15]
Technology
Project Maven uses machine learning algorithms to analyze and fuse vast amounts of surveillance data from multiple sources made possible through data integration using Palantir Technologies.[16][17]
The data sources include photographs, satellite imagery, geolocation data (IP address, geotag, metadata, etc) from communications intercepts, infrared sensors, synthetic-aperture radar, and more. Machine learning systems, including object recognition systems, process the data and identify potential targets, such as enemy tanks or location of new military facility. The training dataset included at least 4 million images of military objects such as warships, labelled by humans. The user interface is called Maven Smart System. It could display information such as aircraft movements, logistics, locations of key personnel, locations on the no-strike list, ships, etc. Yellow-outlined boxes show potential targets. Blue-outlined boxes show friendly forces or no-strike zones. It could also transmit, directly to weapons, a human decision to fire weapons.[17]
Initially the project focused on applications of computer vision. The project's leaders were particularly impressed by model performance on ImageNet.[13] As of 2018, the purpose of the system was AI-enabled analysis of full-motion video. In 2022 it expanded to combatant commands under the AI and Data Acceleration Initiative.[18]
In 2024, it was stated that Maven's key technical contribution was data management: Maven standardizes heterogeneous data through an ontology layer so data can be fused, exchanged across cloud and edge systems, and used by multiple applications. The system was presented as a broader data-centric warfighting system that feeds apps for planning, preparing, and executing operations.[18]
As of 25 March 2025, it was reported that the system at the point incorporated LLMs, generative models, and machine learning, to enhance intelligence fusion and targeting, battlespace awareness and planning, and accelerated decision-making.[15]
Contractors
Initially, The Pentagon collaborated with Google, but in 2018, Google employees, including Meredith Whittaker, staged walkouts protesting Google's involvement in Project Maven.[19][20] The contract was estimated to be worth around $9 million.[21] Subsequently, Google did not renew the contract with Pentagon and Palantir took over the contract.[22][16]
DigitalGlobe provides images and algorithms to Project Maven. The NGA approached IBM for using its AI systems to analyze videos, though IBM did not sign a contract.[23]
Companies that have contributed to the data integration include Palantir, Amazon Web Services, ECS Federal, L3Harris Technologies, Maxar Technologies, Microsoft and Sierra Nevada Corporation. ECS Federal has served as primary support contractor and led AI integration for Project Maven since 2017.[24][25][26] At least 21 private companies had been involved.[27]
Anduril Industries entered the program in 2018 to deploy its sensor fusion platform and edge hardware for data capture.[28] In December 2024 Anduril and Palantir announced a consortium that links Anduril's Lattice Mesh with Maven Smart System and Palantir's AI Platform to move tactical sensor data into AI-supported analyst workflows.[29][30]
In May 2025, the Pentagon raised its contract ceiling for Maven Smart System to $1.3 billion through 2029, up from the previous $480 million. NGA also signed a new $28 million contract to expand access.[31]
As of September 2025, the director of the NGA claimed that by June 2026, Maven will begin to transmit "100 percent machine-generated" intelligence to combatant commanders using LLM technology. Booz Allen was the defense contractor responsible for the LLM-integration phase.[32]
As of 2026 February, Maven was running on Amazon Web Services (AWS), and incorporates a version of Claude, a series of AI systems developed by Anthropic.[33]
In 2013, the intelligence community partnered with Amazon to allow classified material to be processed on AWS. In 2023 and 2024, Amazon invested in Anthropic. Also in 2024, Palantir partnered with Anthropic to integrate Claude into Palantir's software. Anthropic stated that Claude was used to help the military to process data and make decisions.[34] In late 2024, the Pentagon began to integrate Claude into Maven.[35] As reported in 2024 November, Claude 3 and 3.5 family of models were integrated into Palantir's AI Platform running on AWS. They received Defense Information Systems Agency Impact Level 6 accreditation.[36] In 2025 June, Anthropic announced "Claude Gov", which runs in classified environments on the AWS by the national security community. In 2025 July, Anthropic announced that the Department of Defense, through Chief Digital and Artificial Intelligence Office, awarded Anthropic a two-year prototype other transaction agreement with a $200 million ceiling to "prototype frontier AI capabilities that advance U.S. national security".[37][38] In February 2026, Jack Shanahan stated that Anthropic is a partner of Project Maven.[39]
Separately, in April 2025, Anthropic joined Palantir's FedStart program so Claude could be deployed for government customers at FedRAMP High and DoD Impact Level 5, running on Google Cloud, offering assistance to millions of federal government employees.[40]
On 2026 March 4, the Department of Defense designated Anthropic as a supply chain risk, decreeing that all use of Anthropic products to be phased out in all US military within 6 months.[41][42]
Applications
Training exercises
The 18th Airborne Corps is the main tester of Project Maven. With collaborating arms organization in US and UK, it has used Maven and weapons systems connected to it to strike targets from bombers, fighter jets and drones.[17]
Beginning in 2020, Maven was used for live-fire exercises ("Scarlet Dragon exercises").[27] The first took place at Fort Bragg. The test teamed soldiers from the XVIII Airborne Corps with Marines from II Marine Expeditionary Force. An AI system identified a tank in satellite images, the human approved, and the AI system signaled an M142 HIMARS to strike the target (in this case, a tank variously described as either inflated or decommissioned). It was the first AI-enabled artillery strike in the US army.[17] However, the targeting process took 743 minutes.[18][43]
It was also used during Project Convergence 2020 training rounds, as one of several AI systems that participated.[44]
The Scarlet Dragon exercises were overseen by former XVIII Airborne Corps commander Michael Kurilla, who later became the commander of CENTCOM and installed a data-centric command approach that included Maven.[43] In 2024, it was reported that the Scarlet Dragon exercises were still ongoing in 2024, and was planned throughout 2025.[43]
There are 6 steps in the kill chain: identify, locate, filter down to the lawful valid targets, prioritize, assign them to firing units, and fire. In 2024, it was reported that of these 6, Maven can perform 4.[27] A senior targeting officer estimates that with Maven, he could decide on 80 targets per hour, vs 30 targets per hour without Maven.[17] The time for the targeting process has decreased to under 1 minute.[43] The efficiency was comparable with the targeting cell used during Operation Iraqi Freedom, but whereas the OIF used a targeting cell with roughly 2000 staff, the 18th Airborne used a targeting cell with 20 people.[27]
It was stated in 2024 that, while deployed in Europe, Maven iterated with industry through 62 capability evolutions in 10 months.[18]
By August 2025, NATO's Joint Warfare Centre (JWC) said MSS NATO had been deployed across Allied Command Operations, that JWC staff were being trained on it, and that it had already been incorporated into NATO exercises such as STEADFAST DETERRENCE 2025 May, with broader use planned in 2026. JWC described it as NATO's first AI-enabled warfighting command-and-control system.[45] It was later used at STEADFAST DUEL 2025 October as the primary platform for warfighting integration.[46]
In November 2025, the Maven Common Operating Picture was used in SETAF-AF's Lion Deployment Readiness Exercise.[47]
Real use
The first real use of Maven was in 2017 in East Africa, by Special Forces pursuing the Islamist group al-Shabaab.[13]
In early December of 2017, Maven was fielded to defense intelligence analysts to support drone missions against ISIS.[4]
In the 2021 Kabul airlift, Maven was used to display the situation on the ground. It could simultaneously display data feeds, such as aircraft movements, logistics, threats and locations of key personnel such as Chris Donahue.[17]
In the 2022 Russian invasion of Ukraine, the US has used satellite intelligence and Maven Smart System to supply the locations of Russian equipment to Ukrainian forces.[17]
In February 2024, Maven was used for narrowing targets for airstrikes in Iraq and Syria, and was involved in more than 85 airstrikes. It was also used for locating rocket launchers in Yemen and surface vessels in the Red Sea, some of which were destroyed in February 2024 according to CENTCOM.[17][48]
During Hurricane Helene in 2024. The public affairs office of XVIII Airborne Corps used Maven to visualize social media sentiment from Sprinklr.[49]
It was claimed in 2024 that CENTCOM used it as a shared data/AI system that fused 179 live data feeds and supported command-and-control, fires, force protection, and sustainment workflows. It was used in four-star headquarters with over 20 subordinate headquarters. CENTCOM had ~13,000 Maven accounts, of which ~2,500 were regular users who logged in at least a few times a week.[50][18]
It was reported in May 2025 that Maven had over 20,000 active users.[31]
In January 2026, Maven and Claude were used during the intervention in Venezuela.[33][51]
In late February and early March 2026, Maven with Claude was used in the planning and operation of Operation Epic Fury. During planning, the system suggested hundreds of targets, found precise location coordinates, and prioritized those targets. It was claimed to have turned "weeks-long battle planning into real-time operations". During operation, it was used to evaluate a strike after it was initiated.[35][52]