ORCID

Abstract

Contemporary maritime operations such as shipping are a vital component constituting global trade and defence. The evolutiontowards maritime autonomous systems, often providing significant benefits (e.g., cost, physical safety), requires the utilisation ofartificial intelligence (AI) to automate the function of a conventional crew. However, unsecured AI systems can be plagued withvulnerabilities naturally inherent within complex AI models. The adversarial AI threat, primarily only evaluated in a laboratoryenvironment, increases the likelihood of strategic adversarial exploitation and attacks on mission-critical AI, including maritimeautonomous systems. This work evaluates AI threats to maritime autonomous systems in situ. The results show that multipleattacks can be used against real-world maritime autonomous systems with a range of lethality. However, the effects of AI attacksvary in a dynamic and complex environment from that proposed in lower entropy laboratory environments. We propose a set ofadversarial test examples and demonstrate their use specifically in the marine environment. The results of this paper highlightsecurity risks and deliver a set of principles to mitigate threats to AI, throughout the AI lifecycle, in an evolving threat landscape.

Publication Date

2023-01-01

Publication Title

AI, Computer Science and Robotics Technology

Acceptance Date

2023-03-22

Deposit Date

2023-03-23

Embargo Period

2023-04-28

Download

Share

COinS