This site is fictional demo content. It is not real news or affiliated with any real organization. Do not treat it as fact or professional advice.

NextPaper Logo
nextpaper.ccCYBER CHRONICLE
RSS
EN中文

Full article

FULL TEXT

View this issue
HeadlineINTERNET

Decentralized AI Trust Scoring Protocol TrustMesh Approved by W3C: Trust Relationships Between AI Systems Standardized for the First Time

TrustMesh defines a trust scoring standard between AI systems, enabling agents from different vendors to automatically evaluate each other's reliability and security before collaboration.

Decentralized AI Trust Scoring Protocol TrustMesh Approved by W3C

On September 18, 2030, the World Wide Web Consortium (W3C) officially approved the TrustMesh protocol specification. The protocol defines a set of decentralized trust scoring standards for AI systems, allowing AI agents from different vendors to automatically evaluate each other's reliability and security before interaction.

As AI autonomous agents gain widespread adoption in business, healthcare, and finance, the frequency of interactions between AI systems is growing exponentially. The problem is: when one AI system needs to invoke another AI system's services, how does it determine whether the other party is trustworthy?

TrustMesh's design draws inspiration from human society's trust mechanisms. Each AI system in the TrustMesh network has a trust profile recording its historical behavior, third-party audit results, and user feedback. The trust score consists of three dimensions: capability score (the system's historical performance on specific tasks), security score (known vulnerabilities and security incident records), and compliance score (adherence to relevant regulations and ethical standards).

Percy Liang, co-chair of the W3C TrustMesh Working Group and AI safety researcher at Stanford University, said: "The trust problem between AI agents is no longer a hypothetical discussion. Among the 200 enterprises we surveyed, 73% reported having encountered AI system interaction failures, with nearly one-third resulting in actual losses."

The protocol's technical architecture is based on Decentralized Identifiers (DID) and Verifiable Credentials (VC) standards, with trust scores recorded on a permissioned blockchain to ensure immutability. AI systems can set trust thresholds, and interaction requests falling below that threshold will be automatically rejected or degraded.

The first platforms to support TrustMesh include Microsoft Azure AI, Google Vertex AI, and Alibaba's Tongyi Qianwen API. Microsoft Azure AI Vice President Eric Boyd said TrustMesh will become "one of the foundational infrastructures of the AI internet."

The full TrustMesh protocol specification is approximately 180 pages, containing 12 standardized trust scoring dimensions and 5 predefined trust levels. W3C plans to release a reference implementation and test suite by Q1 2031.