The Dutch Navy Is Testing Crewless Warships Run by AI
Two unmanned patrol vessels are circling target ships off the Dutch coast. No sailors onboard. A computer decides where they go.
Key points
- The Royal Netherlands Navy is running a five-week live exercise off Den Helder testing AI-controlled unmanned surface vessels in 2025.
- Two vessels, Defender 1 and Defender 2, patrol around target ships with no crew onboard, guided entirely by computer systems.
- Captain Sjoerd Feenstra, head of the navy's unmanned-systems expertise centre, is leading the trial.
- The stated goal is to remove sailors from situations the navy considers too dangerous for humans.
Two black vessels cut through the North Sea. No crew, no one at the wheel, just sensors and a computer deciding where to go next.
Defender 1 and Defender 2 are unmanned surface vessels, boats that operate without any human onboard, circling target ships in live exercises off Den Helder in the northern Netherlands. The Guardian reported the details of the five-week trial.
Captain Sjoerd Feenstra leads the Royal Netherlands Navy's expertise centre for unmanned systems. He's described the mission plainly: keep sailors out of danger zones. That's the honest pitch. Not spectacle, not novelty. Risk reduction.
What does this mean for the sailors who would normally do this work?
For now, they stay onshore while the computer watches. The Defenders track a target vessel from both sides at a safe standoff distance, a role that traditionally puts a crew in harm's way. An unmanned vessel absorbs that risk instead.
This isn't a finished product, though. A five-week test off a domestic coastline is still a controlled exercise. The navy's probing limits, not declaring victory. Trials like this rarely become frontline capability without years of further development, regulatory approval and integration with existing fleet doctrine, the rules that govern how a navy actually fights.
The broader investment is real. Militaries across Europe are pouring money into unmanned systems at sea, and the pace has quickened. AI2Day covered the first confirmed US combat use of explosive drone boats on 14 July 2026, when three unmanned sea drones destroyed an Iranian submarine and a naval maintenance facility in a single overnight operation.
Feenstra's team is now testing what vessels like these can do autonomously, without a human giving each instruction, and where they still need a person in the loop. The distinction matters enormously: a ship that scouts independently is one thing, a ship that decides to engage is quite another, and the Dutch trial hasn't crossed that line.
Should you worry?
For ordinary citizens, the practical shift is this: navies are beginning to treat some dangerous patrol tasks the way factories treated hazardous assembly work. A machine takes the risky shift; the human steps back. Whether that makes maritime conflict less deadly overall, or simply relocates who bears the risk, isn't a question the technology can settle on its own.
What I'd watch is whether this exercise produces published performance data. Company claims about autonomous vessels are easy to make. Peer-reviewed results from a live naval trial would be something else entirely.
