Drone Subs Can Shoot Now. Who Shoots Back?

Week 42 | September 2026

At the British Underwater Test and Evaluation Centre off Scotland earlier this month, a US Navy Mk 48 heavyweight torpedo dropped away from the Royal Navy's XV Excalibur and ran. Excalibur has no crew. It's a 12-metre experimental XLUUV, and the torpedo was slung underneath it because, reportedly, the weapon wouldn’t fit in the vehicle’s bay. It was the first torpedo either navy had fired from an uncrewed submarine. From approval in February, launch took under seven months.

Media coverage read it as an interchangeability win for AUKUS, and that it is. That was how Admiral Daryl Caudle, the US Chief of Naval Operations, framed it.

I read it from the other end. An autonomous submarine that can shoot is also something an adversary will eventually build and park off someone else's port. So what has anyone built to stop one?

Quite a lot, it turns out, from prototypes to products on sale. None of it sits on a program of record yet.

Excalibur won't be the last. The follow-on, Project Cutlass, moves the work to an Anduril Dive-XL, the design family behind the Royal Australian Navy's Ghost Shark, to push payload integration and tactical employment further.

Ukraine got to the other half of the problem first. Global Mark's Sea Trident ST-1000, shown at Eurosatory in June, is a ten-metre, ten-tonne steel-hulled XLUUV with a one-tonne payload. Its advertised mission list includes intercepting and neutralising other underwater drones. It's the first XLUUV I'm aware of marketed openly as a UUV hunter. Global Mark says it's in the final stage of testing, but there's no public contract, production run or combat use yet.

China has at least five XLUUV designs identified in open-source imagery, with larger types in testing, and this month a submarine-drone mothership was imaged at a yard in Shanghai.

So the threat hunts, and now it shoots too.

In June, the US Navy's Lanternfish 2026 exercise ran a port-protection scenario against medium and large UUVs. Ultra Maritime's Sea Spear sonar, paired with its SSQ-125B active sonobuoy, detected and tracked the targets. Anduril's Seabed Sentry ingested those tracks, and its Lattice software pushed the picture into the Navy's counter-uncrewed command system. Anduril says no human relayed anything between detection and alert. Nothing was defeated; the demonstration stopped there. 

ThayerMahan launched SeaGuard in March, a non-kinetic UUV defeat system designed to pair with its TransparenSea sensing. The company claims TRL 8+ and says it's open for orders.

Then in July, Lockheed Martin agreed to buy Ultra Maritime for US$3.45 billion. Once the deal closes, the sonar at the centre of Lanternfish will belong to a prime.

My read: the sonar is the easier half. The prize is software that fuses someone else's sensors into an existing kill chain.

The most candid official statement came from the US Coast Guard. Its 24 July sources-sought notice for counter-UUV systems says it plainly: there are "effective technologies with the capability to detect and track UUVs, but few tested and fielded defeat/mitigation systems."

That's a market survey. It asks industry what exists and commits no money to buying any of it.

Look further and there's activity, just earlier-stage than you'd want. In March, DIU and UK Defence Innovation launched REEF, a joint counter-UUV prototyping effort. In July, the Navy's Strategic Systems Programs office asked industry for prototypes to detect and defeat uncrewed systems around its missile-submarine bases. At BALTOPS in June, NSWC Panama City tested air curtains, nets and neutralisation charges against UUVs. What I can't find, in the US, UK, Australia or NATO, is a public counter-UUV program of record. Even the AUKUS Pillar 2 UUV payload project announced in May doesn't name counter-UUV among its mission sets.

I should caveat one thing. A requirement can sit behind classification, and so can the money. US budget books carry classified lines, and Australia's IIP withholds some projects. My claim is about the public record, and what stands out there is the asymmetry: the vehicles that hunt are programs of record, while the counter to them is still at the prototype and exercise stage. With armed UUVs heading into service, that should make someone uncomfortable. 

Suppose the sensor works. A contact is logged, submerged, a few miles off a naval base. Now what?

Detection alone is hard. Small electric UUVs are quiet and harbours are loud. The binding problem, though, is classification: hostile vehicle or a university's research glider? Every false positive costs something.

The law is harder, and in some ways more interesting than the sonar.

UNCLOS Article 20 requires submarines and "other underwater vehicles" to navigate on the surface and show their flag in the territorial sea. A submerged, unflagged UUV fails that test, so it isn't in innocent passage (assuming it ever had a right to it, which UNCLOS never settles), and Article 25 lets the coastal state take the "necessary steps" to prevent it. UNCLOS doesn't say what those steps include.

Destroying it is where things get difficult. Armed force against an unknown submerged contact is lawful only in self-defence under Article 51 of the UN Charter, and presence alone doesn't amount to an armed attack. Lieutenant Colonel Brent Stricker at the US Naval War College's Stockton Center makes exactly that point.

Status is messier again. An uncrewed vehicle probably fails the Article 29 definition of a warship, which assumes a commander and crew, yet it may still be sovereign-immune state property. China's December 2016 seizure of a US Navy glider, just as USNS Bowditch was about to recover it, showed how fast that turns into a diplomatic fight. Beijing handed it back five days later.

Outside wartime, state practice so far is quiet recovery. Fishermen in Indonesia and the Philippines have repeatedly hauled up Chinese underwater drones, mostly gliders, since 2019 and handed them to the authorities for examination. Nobody fired at anything. Where UUVs have been attacked, by US forces against the Houthis in 2024 and by Israel against Hamas in 2021, the attacker was known and hostilities were already under way.

Attribution sits under all of it. Force against a UUV you can't attribute is very hard to justify, and a vehicle built to be deniable is built to defeat attribution. No state has squarely answered this. Sweden comes closest. Its published rules let the navy use force without warning against a foreign submarine submerged in its internal waters, whoever owns it. The ordinance just says “submarine.” It doesn’t say whether one needs a crew, and as far as I can find, neither does anyone else.

The torpedo fired off Scotland was reportedly a Mk 48 Mod 7, the variant the US and Australia developed together. RAN operators took part in Lanternfish, and the design family behind Ghost Shark is next in line for allied payload-integration work. There's no public Australian counter-UUV line. As the Week 26 brief laid out, the country is also down to two ageing minehunters, with SEA 1905 pushed to the early 2030s. Stirling and Port Hedland are still the ports at risk.

The allies have shown a drone submarine can fire a torpedo. Nobody has said publicly what you're allowed to do when one turns up off your harbour.

The MASS Code says a master stays responsible for an autonomous vessel, even from ashore. It sets goals for connectivity and cyber resilience, but says far less about how the link to that master should be secured and tested. Class societies are certifying hulls and autonomy stacks, and vessels are being delivered while that question stays open.

Since you have been, thanks for reading.

Cheers,

Mick

Ocean Tech Intelligence provides informational analysis only. Nothing in this publication constitutes financial, investment, legal, or strategic advice. Readers act on this content at their own risk. For full details see our Disclaimer.