The UK and US are opening another route for joint development of military artificial intelligence and autonomous systems after a British-built uncrewed submarine fired a heavyweight torpedo during an AUKUS trial in UK waters.

The proposed partnership will connect the UK Ministry of Defence Rapid AI Delivery Taskforce with its US counterpart responsible for digital and artificial-intelligence capability. The governments say the work will focus on trusted and interoperable systems for areas including critical-infrastructure protection, surveillance, deterrence, and autonomous operations.

The announcement follows a live firing involving Excalibur, the 12-metre extra-large uncrewed underwater vehicle developed in Plymouth through Project Cetus. The UK government says a heavyweight torpedo jointly developed by the US and Australia was launched from Excalibur in UK waters, describing the event as the first such firing by either the UK or US from an uncrewed underwater vessel.

The claim comes from the government announcement and should be understood as its description of the test rather than an independent assessment of combat capability. The release also does not state that lethal targeting decisions were delegated to the vehicle itself.

AUKUS moves from experiments towards integration

Excalibur has been used as a testbed since entering trials in 2025. The Submarine Delivery Agency transferred the vehicle to the Royal Navy after a programme that included remote operation from Australia while the vessel remained in UK waters, alongside tests involving quantum navigation technology.

The platform forms part of AUKUS Pillar 2, under which Australia, the UK, and US work on advanced technologies including autonomy, AI, electronic warfare, quantum systems, and undersea capability.

Interoperability is one of the difficult engineering problems because an autonomous vessel becomes more useful to allied forces when command systems, communications, payload interfaces, and operational procedures allow equipment from several countries to work together without bespoke integration for every mission.

The new UK-US arrangement is intended to connect teams responsible for that work, although the announcement does not yet set out a budget, procurement programme, or delivery timetable for the partnership itself.

Autonomy changes defence procurement

Military autonomy can allow uncrewed systems to operate where deploying people is dangerous, expensive, or operationally difficult. Undersea platforms can perform sensing, surveillance, seabed, and other missions without every vessel carrying a crew.

Software-driven capability also complicates procurement because armed forces are buying systems whose behaviour depends on software, data, sensors, communications, and integration rather than a fixed mechanical specification alone.

Weapons integration raises the assurance threshold further. A vehicle able to carry and launch a weapon is not automatically making autonomous weapons decisions, and the distinction should remain explicit when AI, autonomy, and a live firing appear in the same programme.

Command authority, rules of engagement, communications loss, human supervision, cybersecurity, and behaviour under unexpected conditions remain separate operational questions even when the underlying platform performs well.

Plymouth becomes part of the supply chain

Excalibur was developed with Plymouth-based MSubs, connecting the programme to a growing maritime-autonomy cluster in the south-west of England. Autonomous maritime systems also draw on sensors, secure communications, navigation, software assurance, simulation, power systems, payload integration, and specialist engineering.

That can create industrial opportunities beyond the platform manufacturer, although moving from demonstrators into sustained procurement remains a separate hurdle. Defence organisations have often proved capable of funding trials more quickly than they can change doctrine, certification, budgets, and legacy platforms.

The torpedo firing provides another interoperability test and the bilateral partnership creates an organisational route for further work, but neither settles how autonomous systems will ultimately be deployed or how much authority they will receive.

The remaining distance between experimentation and routine military use will be determined as much by assurance, procurement, command structures, and operating rules as by further advances in AI itself.