What is it?
Project Maven is the Pentagon’s lead artificial intelligence program, initiated in 2017 as a targeted experiment to aid military analysts in processing the overwhelming amount of drone footage from conflict zones.
Operators were overwhelmed with images, scrutinizing frame by frame for fleeting objects of interest. Maven was designed to locate the needle in the haystack.
Eight years later, the program has transformed into a far more extensive system: an AI-enhanced targeting and battlefield management apparatus that has significantly expedited the kill chain—the sequence from detection to destruction in warfare.
How does it work?
Maven operates like the air traffic control of warfare and its cockpit.
Aalok Mehta, director of the CSIS Wadhwani AI Centre, described it as “essentially an overlay” that integrates sensor data, enemy troop intelligence, satellite imagery, and troop deployment information.
In practice, this involves swiftly analyzing satellite feeds to track troop movements or identify targets, while also capturing a “snapshot of the operational theater” to devise optimal strategies for targeting specific objectives.
A recent online demonstration featured a Pentagon official explaining how Maven “magically” converts an identified threat into a targeting workflow, evaluating available assets and providing commanders with options.
The advent of ChatGPT has represented another significant advancement, expanding the technology’s accessibility to a broader user base who can engage with Maven using natural language.
Currently, this functionality is powered by Anthropic’s Claude—though that partnership is ending sourly after the Pentagon objected to the AI lab’s stipulation against using its model for fully automated strikes or monitoring US citizens.
Why did Google say no?
The ethical implications were a significant concern in the early days of Maven, during which Google served as the program’s initial AI contractor.
In 2018, over 3,000 employees signed an open letter protesting the company’s involvement, claiming the contract crossed an ethical boundary. Numerous engineers resigned as a result.
Google chose not to renew the contract upon expiration and subsequently established AI principles explicitly prohibiting participation in weapons systems.
This episode highlighted a divide in Silicon Valley between engineers who considered autonomous targeting an ethical violation and defense officials who deemed it crucial.
More recently, Google lifted its AI policy restrictions and indicated a stronger focus on national security efforts. The Pentagon has mentioned that Google, along with xAI and OpenAI, are contenders to replace Claude in the Maven program.
What is Palantir’s role?
In 2024, Palantir—partially funded by the CIA and designed for government intelligence work—entered the vacuum left by Google.
The firm has reportedly become Maven’s primary tech contractor, forming the operational backbone of the program.
Palantir CEO Alex Karp clearly articulated the stakes.
“This is a have, have-not world,” he asserted at a recent Palantir event, emphasizing the necessity for the West to develop capabilities that others lack.
A system that reduces the kill chain time from hours to seconds renders adversaries obsolete, he noted.
How has it fared?
Both the Pentagon and Palantir have refrained from commenting on Maven’s effectiveness in the current conflict with Iran.
US operations have been conducted at a sustained pace, and it is reasonable to assume that Maven’s capacity to expedite targeting and firing processes has been central to these efforts.
The Centre for Strategic and International Studies reports that after three weeks, the US strike campaign stabilized at a rate of 300 to 500 targets per day.
In the first 24 hours of Operation Epic Fury, US forces targeted over 1,000 locations, which included a school in a facility previously used as a military base, according to various news reports. Iran has claimed the attack resulted in the deaths of 168 children aged seven to 12 and left many others injured.
