Decentralised Governance-Driven Architecture for Designing Foundation Model based Systems: Exploring the Role of Blockchain in Responsible AI

Foundation models lack clear governance structures: The paper identifies eight specific governance challenges across three critical dimensions—decision rights (who makes choices), incentives (what motivates behavior), and accountability (who is responsible)—highlighting that current AI systems often operate without adequate oversight mechanisms to ensure trustworthiness and prevent misuse.

Blockchain offers a potential solution for distributed oversight: Rather than relying on centralized control, the authors propose using blockchain technology to create a decentralized ledger system that can track decisions, actions, and accountability across foundation model-based AI systems, distributing governance responsibilities rather than concentrating them in single organizations.

Current governance gaps threaten multiple stakeholders: The paper emphasizes that inadequate governance of foundation models poses risks to humans, society, and the environment—suggesting that without proper oversight structures, these powerful AI systems could cause widespread harm across multiple domains.

Decentralized governance could improve transparency and trust: By leveraging blockchain's distributed nature, the proposed architecture aims to make governance decisions and processes more transparent and verifiable, potentially increasing public confidence in how foundation models are developed, deployed, and monitored.

Governance requires multi-dimensional oversight: The paper's focus on three governance dimensions (decision rights, incentives, and accountability) suggests that responsible AI requires coordinated attention to who decides, what motivates behavior, and who answers for outcomes—not just technical safeguards alone.