Marine AI has been awarded new funding through the UK Defence and Security Accelerator (DASA) to advance the use of large language models (LLMs) in maritime autonomy. According to the company’s official announcement, the project focuses on improving how autonomous vessels communicate and make decisions at sea.

Marine AI explained that its technology has already proven capable of generating spoken VHF transmissions that adhere to the International Maritime Organization’s Standard Marine Communication Phrases (SMCP). This early-stage capability has been linked to the company’s GuardianAI navigation system, which has been developed under initiatives including DARPA’s NOMARS and the Royal Navy’s CETUS program, and tested on the Royal Navy’s vessel XV Patrick Blackett.

With the newly awarded support, Marine AI plans to refine the system further. The company noted that upcoming work will involve collecting passive VHF data, gathering both guided and real-world operational datasets, and advancing noise reduction models. These improvements are expected to deliver a stronger and more capable LLM, which will be trained and demonstrated during the project timeline.

According to Marine AI, the ultimate goal is to provide autonomous surface vessels with more reliable communication and decision-making tools, ensuring safer navigation and more effective performance in complex maritime environments.