Algorithmic warfare

Algorithmic warfare refers to the integration of artificial intelligence (AI), machine learning, and large-scale data analytics into the conduct of military operations. Rather than describing a single weapon system, the term captures a broader shift in doctrine: compressing the observe–orient–decide–act (OODA) loop by delegating parts of sensing, analysis, and targeting to algorithms.

The concept gained institutional traction with the U.S. Department of Defense's Project Maven, established in April 2017 under the Algorithmic Warfare Cross-Functional Team. Maven applied computer vision to full-motion video collected by drones, automating object detection that had previously consumed thousands of analyst hours. The project became publicly contentious in 2018 when Google declined to renew its contract following internal employee protests.

Typical components include:

• Sensor fusion across satellites, drones, signals intelligence, and open-source data.
• Automated target recognition (ATR) to identify vehicles, personnel, or infrastructure.
• Decision-support systems that recommend courses of action to human commanders.
• Autonomous and semi-autonomous platforms, including loitering munitions.

Real-world deployments have been documented in the Russia–Ukraine war, where systems such as Palantir's analytics platforms and Ukrainian targeting software (reportedly including GIS Arta) have shortened strike cycles. Reporting by +972 Magazine and The Guardian in 2023–2024 described Israeli systems known as Lavender and The Gospel used to generate targeting recommendations in Gaza, intensifying debate over algorithmic accountability.

Legal and ethical concerns center on compliance with international humanitarian law (IHL)—particularly the principles of distinction, proportionality, and precaution under Additional Protocol I to the Geneva Conventions. The UN Group of Governmental Experts on Lethal Autonomous Weapons Systems (GGE on LAWS), convened under the Convention on Certain Conventional Weapons since 2017, has debated whether meaningful human control can be preserved when targeting is algorithmically mediated, though no binding instrument has been adopted.