TrustChain-Review: Blockchain and game theory for trustworthy AI code review

TrustChain-Review, introduced by Mohammad Naserameri in arXiv:2607.27310, tackles a growing problem: AI-assisted software development speeds up coding and review but undermines accountability. Developers may submit under-verified code, reviewers can approve changes with minimal inspection, and centralized reputation records are hard to audit. The framework layers a blockchain-based evidence system on top of a three-player game-theoretic model involving developers, reviewers, and the platform. A risk-adaptive rule applies stronger governance only when the expected benefit justifies the cost, preventing over-regulation on routine changes.

In a controlled simulation calibrated with the Diff Quality Estimation dataset, Naserameri compared six governance configurations over 30 independent runs. The full-evidence configuration delivered the strongest reputation accuracy, trust convergence, and detection rates for malicious and superficial reviews, but it also carried the highest governance cost. The risk-adaptive configuration cut that cost significantly and improved cost-efficiency by applying strict controls selectively, with modestly lower trust and detection results. The conclusion: high-stakes, audit-sensitive changes warrant full-evidence governance, while routine contributions benefit from a more practical, cost-effective selective approach. The paper spans 24 pages with 3 figures and 10 tables.