Trusta.AI’s most useful decomposition is into three separate layers: TrustScan for sybil detection, TrustGo for value/reputation scoring, and TAS for publishing attestations. That is much more analytically useful than treating it as a generic proof of humanity vendor.
The sybil-detection materials are unusually explicit about method. Trusta says it first runs community detection over asset-transfer graphs and gas-provision graphs, then refines suspicious clusters with K-means-like behavioral screening over transaction-history and profile variables.
The MEDIA score is not an opaque one-word reputation badge in the docs; it is defined as a weighted score across Monetary, Engagement, Diversity, Identity, and Age, with a tunable sigmoid transformation used before aggregation.
TAS is structurally important because it makes the attestation control plane visible. Portals define issuance entrypoints and verification routing, schemas define field layouts, modules add verification or side effects, and the attestation registry stores the outputs.
The docs explicitly compare TAS to EAS and Verax, which helps place Trusta in the attestation-registry lineage rather than filing it only as an anti-sybil company.
Test of Humanity adds a second interesting control surface: personalized knowledge-based authentication based on wallet history, with answer correctness, response time, and answer patterns all contributing to the judgment.
The broader whitepaper language matters because Trusta is not limiting itself to human-uniqueness claims. It explicitly frames the roadmap as moving from Proof of Humanity toward Proof of Intelligence, trying to distinguish humans, bots, and AI agents.
Specific source signals worth retaining
From POH to POI is the best source for Trusta’s two-phase sybil framework, suspicious graph motifs, TON-specific Test of Humanity, and the emerging AI-agent identity framing.
Reputation & Credit Assessment is the best source for the exact MEDIA dimensions, point weights, and sigmoid-based aggregation logic.
Trusta Attestation Service is the best source for the registry architecture and for the portal / schema / module / attestation decomposition.
Introduction to Sybil Score and MEDIA Score is the cleanest concise statement of how TrustScan and TrustGo relate to each other operationally.
The homepage and product pages are useful mainly for confirming the current packaging: Trusta is presenting itself as a combined identity, scoring, and attestation network rather than a single API feature.
Why this entry belongs in the corpus
It strengthens the identity-and-reputation cluster by keeping anti-sybil detection, account-value scoring, and attestation issuance distinct.
It is a useful comparison point for Human Passport, Orange, Nomis, Humanity Protocol, World ID, EAS, and Verax because it exposes where practical control would sit: graph construction, scoring policy, question generation, issuer rules, and attestation acceptance.
The durable insight is that many proof of humanity products are really layered systems with at least four separate control planes: suspicious-account detection, score or classification design, attestation issuance, and downstream relying-party interpretation. Trusta keeps those seams visible enough to study.