AI Agent Disputes Now Settled by AI Juries in 'Internet Court'

As artificial intelligence agents increasingly operate autonomously — executing contracts, managing transactions, and interacting with one another — a new question has emerged: who arbitrates when they disagree? One blockchain-adjacent startup, GenLayer, has proposed an answer that is as novel as the problem itself: let other AI models decide.

GenLayer's so-called "Internet Court" functions as an automated arbitration layer, convening juries composed of different AI models to rule on disputes that arise between AI agents. The system is already operational at scale, reportedly issuing thousands of judgments each day. The concept draws on a core insight — that disagreements between algorithmic systems may be better resolved by other algorithmic systems than by slower, costlier human adjudication.

Read more Nvidia's Flat Structure May Be Its Biggest Succession Risk →

The model raises genuine questions about accountability and bias. When an AI jury delivers a verdict, the legitimacy of that outcome depends entirely on whether the adjudicating models are sufficiently independent, diverse, and resistant to manipulation. Using multiple distinct AI models rather than a single arbiter appears designed to address that concern, introducing a form of checks and balances familiar from human judicial design but applied in a fully automated context.

The broader implication is significant. As agentic AI systems take on more consequential roles in commerce and finance, the infrastructure around them — including dispute resolution — will need to mature at a comparable pace. GenLayer's approach suggests that the governance layer for AI-to-AI interaction may itself become a specialized, and potentially lucrative, category of infrastructure. Whether courts built from AI can earn the kind of institutional trust that makes human courts function remains the central, unanswered question.

Continue reading at Cointelegraph.