In a dark room in London's Shoreditch district, a giant screen displaying a map of the Baltic Sea flashes to life. With a few keystrokes, maritime navigation data is overlaid on the map, showing ships on the move. Another command displays the route of an undersea communications cable.
One of them soon appears as a red icon, a Liberian-flagged ship bound for the BCS East-West Interlink internet cable between Lithuania and Sweden.
Based on the vessel's profile data, dominAI, an AI-powered software that monitors hypothetical situations, identified the vessel as a potential threat. Operators can immediately dispatch military drones or naval vessels to intercept. The system then runs simulations of each possible response and assigns a probability of success to guide decision-making.
This scenario is a demonstration based on real-world events. In November 2024, East-West Interlink had one of its two undersea communication cables damaged within 24 hours by an unknown attacker. Authorities suspected sabotage.
DominAI is the brainchild of Hadean, a London-based former gaming company that is now positioning its software as the command and control solution needed by the British military for the next generation of conflict.
The UK has earmarked £1 billion to build a 'digital targeting web'. It is an effort to use artificial intelligence to integrate data from multiple sources, both civilian and military, into a single command network. This will enable commanders to make faster, more informed decisions about how to quickly scale up responses from simple surveillance to armed intervention against rapidly emerging threats.
“AI underpins the entire military architecture and is, and will continue to be, central to targeting processes,” said Sir Richard Barrons, former British Army general and one of the authors of the UK's 2025 Strategic Defense Review.
Although fully autonomous, intelligent machines are still years away from appearing on the battlefield, AI is already accelerating the way humans process information and make decisions under pressure. The United States has already introduced a command and control system that utilizes AI. Designed by data intelligence group Palantir, Maven was reportedly used in combat during the 2024 US airstrikes in Yemen. NATO has since adopted a variant of this software.
The first challenge AI is solving is moving data on battlefields without cell towers and overcoming widespread electronic interference, said Will Bryce, co-founder and CEO of Arondite, which developed the smart system called Cobalt.
“A lot of innovation is being done to minimize the amount of data that needs to be moved.” That way, AI models can process large streams of images and sonar feeds, reducing dependence on tired human operators. “From there, it's orchestration: making sure we direct the best assets available, whether it's drones, robots, or teams of humans, and execute them as quickly as possible,” added Bryce.
However, speed is only part of the transformation promised by AI. Another is the concept of military mass, or the idea that autonomous drones and robots can fill numerical gaps in wars against larger adversaries.
NATO's 4,300-kilometre eastern border with Russia is sparking new thinking within the alliance about how to use drones to make up for gaps in numbers. The Ukraine war has already demonstrated a revolution in robotics, with inexpensive, remotely controlled first-person view (FPV) drones helping the Ukrainian military counter the massive Russian military. The Russian military also uses drones.
The next step is autonomy. Drones and ground robots no longer require one-on-one human operators. In November, the U.S. Marine Corps posted a video of its troops training with quadcopter drones manufactured by U.S.-based Orterion. The company says the drone runs software that “applies real-time AI to enable soldiers to deploy swarms that act as a single unit.”
“For a four-person team covering a 20km front line in Ukraine, there is only so much we can control using FPV and robotics. [So] Elements of autonomous response will also be necessary. ” Ukraine recently approved the battlefield deployment of semi-autonomous tracked robot Krampus, an armed unmanned ground vehicle (UGV).
A degree of autonomy is also important for navigation in the face of electronic warfare jamming that can paralyze direct radio control by the operator. German startup ARX Robotics is deploying UGVs in Ukraine that can navigate to the next waypoint while avoiding collisions even when communication is not possible, said David Roberts, chief executive of ARX UK.
In the skies, the UK is investing around £2 billion a year in the Global Fighter Plan, with plans to build a sixth-generation fighter with an “optional crew” by 2042 in collaboration with Japan and Italy. The plan would work with so-called “loyal wingmen,” unmanned aircraft already being developed to fly alongside manned aircraft.
Beneath that wave, autonomous submarines such as Anduril's Ghost Shark, recently sold to Australia, and submarine-detecting underwater robots such as Hellsing's SG-1 Fathom are rapidly becoming a reality.
These durable drones, or “gliders,” move by varying their buoyancy and have been around for decades. However, Hellsing has equipped its system with an AI called . lulato classify the acoustic characteristics of ships and submarines.
But as autonomy expands, ethical and legal issues deepen. Can machines really distinguish between combatants and civilians, or can they grasp the moral weight of so-called collateral damage?
“Given that humans are ultimately responsible for ensuring compliance with international humanitarian law, it is important to maintain human judgment in context,” warns Jessica Dorsey of Utrecht University. She warns that AI systems have the potential to compress complex moral and legal judgments into “algorithmic models.”
Similarly, when machines generate hundreds of potential targets per hour, operators risk “automation bias'' (trusting the system by default) or “action bias'' (feeling forced to act because the system demands it). As Dorsey points out, “The question is not simply whether algorithmic systems are useful, but how they reshape our mental structures and processes of responsible command.”
The extent to which humans delegate decisions to algorithms can be the difference between winning and losing. But this logic could ultimately pose even greater dangers.
