Key points are not available for this paper at this time.
In an increasingly connected world, each device is expected to be linked to the internet and, consequently, to other objects with which it can communicate. As the Internet of Things (IoT) expands, so do social, legal, and ethical concerns. Moreover, IoT systems are evolving to process data and provide recommendations based on their findings, a capability that stems from the integration of Artificial Intelligence (AI) and Machine Learning technologies. The convergence of IoT and AI (AIoT), along with the unique aspects of AIoT architectures that differ from non-IoT-enabled AI systems, necessitates a thorough review of specific considerations for building trustworthy AIoT systems. Ensuring trustworthiness in AIoT is crucial due to the increased complexity and potential vulnerabilities introduced by this convergence. This article introduces TrustAIoT, a structured framework for the development and long-term governance of trustworthy AIoT platforms. The framework integrates ethical, legal, and technical dimensions, and consists of both a multi-layer guideline and a lifecycle-oriented process tailored to the specific architectural characteristics of AIoT systems. Based on a systematic literature review, trustworthiness-related technical, ethical, and legal elements are cross-referenced and contrasted with the operational and architectural needs of AIoT environments, ensuring that all critical aspects are addressed. Challenges in applying these elements across heterogeneous architectures are analyzed, leading to the definition of a guideline for transferring trust principles across the different layers of an AIoT platform, and a process for constructing and maintaining such platforms. A platform is understood as an environment comprising infrastructure, tools, services, processes, and components for developing, deploying, and operating applications. The TrustAIoT framework consolidates these artifacts into a cohesive approach that supports trust-oriented decision-making throughout the platform lifecycle, from architectural design to project deployment. The proposed guideline and process form a unified framework that ensures high trustworthiness standards, thereby enabling the reliable development of multiple projects and applications within AIoT ecosystems.
Braga et al. (Mon,) studied this question.