Trusta.AI

• Name: Trusta.AI
• URL: https://www.trustalabs.ai/
• Category: sybil-resistance and reputation middleware / onchain identity scoring / attestation infrastructure / AI+crypto identity network
• Summary: Trusta.AI is worth cataloging not as just another anti-sybil vendor or wallet-score dashboard, but as a layered identity-control system that keeps three usually-collapsed functions separate: sybil detection (TrustScan), onchain reputation/value scoring (TrustGo and its MEDIA score), and attestable public outputs (TAS, the Trusta Attestation Service). The primary materials describe a two-phase clustering-and-behavior system for identifying coordinated wallets, a five-dimension score for ranking account value, and a portal/schema/module attestation architecture explicitly compared to EAS and Verax. That makes Trusta.AI a useful comparison point for Human Passport, Nomis, Orange Protocol, Humanity Protocol, World ID, and generic attestation stacks: it turns identity verification into a pipeline of graph analysis, score design, attestation issuance, and downstream reuse instead of flattening everything into one vague proof of humanity claim. Keep this note on that layered mechanism surface; use trusta-ai when the sharper question is who authors the models, reviews API access, signs attestation payloads, and packages anti-sybil workflow for downstream projects.
• What it does:
  • Uses TrustScan to score EVM EOAs for sybil likelihood by analyzing asset-transfer graphs and account-behavior patterns
  • Uses TrustGo to compute a 0–100 MEDIA score based on Monetary, Engagement, Diversity, Identity, and Age dimensions of onchain activity
  • Defines TAS (Trusta Attestation Service) as a public blockchain attestation registry with portal, schema, module, and attestation-registry layers
  • Describes issuance flows where a Trusta backend signs score data and a frontend/wallet interaction routes that data through a portal contract for onchain attestation
  • Positions proof outputs as reusable public identity primitives, especially Proof of Humanity and MEDIA reputation attestations
  • Expands the concept beyond human-only verification into Test of Humanity, Proof of AI Agent, and a broader Proof of Intelligence / AI-identity framework
• Key claims:
  • The most useful analytical feature is that Trusta.AI keeps detection, scoring, and attestation as separate layers. TrustScan identifies suspicious wallet clusters, TrustGo scores value or contribution, and TAS publishes reusable attestations. Many competing identity products bundle those layers together.
  • The sybil-detection docs are more explicit than most marketing pages. Trusta says Phase 1 runs community-detection logic over asset-transfer graphs and gas-provision graphs to detect suspicious star, tree, and chain funding patterns; Phase 2 refines those clusters using K-means-like screening on transactional and profile variables.
  • MEDIA is not a generic reputation label; it is a specific weighted score design. The docs say it combines Monetary (25), Engagement (30), Diversity (15), Identity (10), and Age (20) dimensions, with tunable sigmoid normalization before weighted aggregation.
  • TAS is worth retaining because it exposes the attestation control plane directly. Portals define issuance entrypoints and verification chains, schemas define field layouts, modules inject custom verification or side effects, and the registry stores the resulting attestations. That is a clearer decomposition than many flatter credential products.
  • Trusta’s Test of Humanity is a distinct control surface rather than a minor product feature. The docs describe personalized, knowledge-based questions derived from a wallet’s activity, plus response-time and answer-pattern analysis, as a lighter-weight alternative to biometric or document-heavy proof systems.
  • The project also matters as a bridge between identity and AI-agent infrastructure. The whitepaper materials explicitly move from Proof of Humanity toward Proof of Intelligence, arguing that future systems need to distinguish humans, bots, and AI agents rather than only proving human uniqueness.
  • The durable governance insight is that Trusta’s real power would sit in scoring-model design, graph-cleaning assumptions, question-generation policy, portal/module ownership, and issuer acceptance rules — not only in any eventual token or attestation surface.
• Whitepaper: Trusta.AI does not present a single clean standalone PDF in the current primary sources, but the official GitBook whitepaper/docs set is substantive enough to function as the canonical source base for this pass; see ../whitepapers/trusta-ai-primary-sources-2026-05-12.md.
• Sources:
  • https://www.trustalabs.ai/
  • https://trusta-labs.gitbook.io/trustaai
  • https://trusta-labs.gitbook.io/trustaai/from-poh-proof-of-humanity-to-poi-proof-of-intelligence
  • https://trusta-labs.gitbook.io/trustaai/trusta-attestation-service
  • https://trusta-labs.gitbook.io/trustaai/reputation-and-credit-assessment
  • https://trusta-labs.gitbook.io/trustaai/products/trustscan/introduction-to-sybil-score-and-media-score.md
  • https://trusta-labs.gitbook.io/trustaai/products/trustgo.md
  • https://trusta-labs.gitbook.io/trustaai/products/trustscan.md
  • https://github.com/TrustaLabs/

Internal linkages

• Company / operator boundary above the scoring-and-attestation stack: trusta-ai.
• Useful adjacent mechanism reads: world-id, gitcoin-passport, and attestation-station.

Protocol / company boundary

• Keep this project note on the mechanism layer: TrustScan’s detection logic, TrustGo/MEDIA score design, TAS portal/schema/module architecture, and the way those layers compose into one identity pipeline.
• Use trusta-ai for the operator boundary above that stack: API-applicant review, score-model tuning, attestation-data signing, issuer policy, partner consulting, and the productized workflow ecosystems may start treating as eligibility or reputation truth.
• That split matters because how does this identity stack work? and which company decides who gets scored, how the score is exposed, and which attestation flows are operationally trusted? are related but not identical questions.

Control surface

• At the project layer, the leverage sits in the architecture itself: how detection, scoring, and attestation are separated, what data gets measured, and how reusable proof objects are exported downstream.

• The thicker operator dependency now points upward to trusta-ai, where API admission, graph-cleaning assumptions, threshold policy, backend signatures, and partner-facing sybil workflow become easier to analyze as one company boundary.

• Read Trusta.AI here as identity-scoring and attestation middleware, not as a neutral human-registry baseline.

• Last reviewed: 2026-08-11 UTC