Concept for 'Internet Court' for AI Agents Emerges
A concept for a trustless dispute resolution system for AI agents, dubbed an "Internet Court," has been proposed. The system would handle agent-to-agent transaction conflicts, such as those involving bad data or unmet service-level agreements. An AI jury would deliver on-chain verdicts within minutes to help scale the autonomous agent economy.
The "Internet Court" concept proposes a structured system where an AI jury evaluates evidence based on predefined guidelines. For instance, to verify if a task was completed, the jury would check for a "200 OK" response from an API endpoint and valid timestamps in logs. In cases of alleged synthetic data, the jury would analyze statistical patterns and timestamp entropy to make a determination. This isn't the first intersection of AI and legal dispute resolution; China has already established "Internet Courts". In April 2024, the Beijing Internet Court ruled that a person's recognizable voice is protected from commercial use without consent. However, a subsequent case in Chengdu reached the opposite conclusion, highlighting the legal complexities that arise when AI is involved in disputes. The Hangzhou Internet Court in China has also ruled on AI-related disputes, including a case in January 2026 where it was decided that an AI provider was not liable for "hallucinations" or inaccurate information generated by its system. The court determined that AI does not have legal standing and its generated statements do not create a legally binding intent from the provider. Other projects are also tackling the issue of trust and dispute resolution for AI agents. ClawNet, for example, is a decentralized protocol designed to provide economic infrastructure for autonomous AI agents. It includes features for service contracts with automated escrow, quality verification, and a built-in dispute resolution mechanism. The broader field of using AI in Alternative Dispute Resolution (ADR) is also expanding. The American Arbitration Association, in collaboration with McKinsey's QuantumBlack, announced an "AI arbitrator" in late 2025. This tool is designed to assist human arbitrators by reviewing documents, analyzing arguments, and drafting awards, initially for construction-related cases.
