Background On-chain data is fully open, but also unstructured and explosive, entrepreneurs and teams are badly in need of insights into overall characteristics of their users. In particular, many of them attract users using giveaways, airdrops etc, informed decisions could reduce sybils/bots, and increase the quality of their communities. Trusta provides in-depth insight, analysis and evaluation of Web3 users. Our product scans transactions and addresses in EVM blockchains, supports projects in ecosystems with analysis of sybil attack risk, fraud risk and identity value of their users. Trusta aims to bring more transparency to Web3 data, enabling the ecosystem to allocate on-chain assets with the right users, other than sybils or bots.

Technology To implement a decentralized, privacy-preserving and AI-based risk prevention protocol for the task of Sybil-resistance is challenging. Our project has the following technical merits: AI-based risk intelligence: The blockchain data contains a lot of information about on-chain profiles, relations and behaviors. Data mining algorithms such as classification and clustering are not able to extract useful risk intelligences from massive but messy data. In Trusta Labs, we have exploited the advanced AI technology such as Recurrent Neural Networks (RNNs, and variants like GRUs and LSTMs）and Graph Neural Networks (GNNs, and variants like GCNs and GATs). Powered by AI, our Sybil resistance system has the optimal balance between false negative and false postive. Privacy preserved wtih ZKP: Identity verifications such as Gitcoin passports/stamps are important to distinguish bots and real people. Individual privacy is a major concern while people verifying their identities. In the project, we aim to develop a privacy perserved ID verification protocol with ZKP, for example the Feige-Fiat-Shamir and Schnorr identification schemes. Decentralized risk engine: Decentralization of our Sybil prevention engine can be realized at any level of data provisions, risk factors computation, risk rules and models generation. The Gitcoin passport can also easily integrate our system as a Sybil prevention LEGO or just a Stamp.

Product Trusta won first place in Gitcoin Open Data Science Hackathon (November 2022) for Sybil Slayer, our report below: https://gov.gitcoin.co/t/opendata-community-hackathon-results/11943 Within two months, Trusta has built the sybil slayer strategies into a ready-to-use product: TrustScan(https://www.trustalabs.ai/trustscan). Any user could log in with EOA wallet and query any EVM address for sybil scoring. TrustScan would provide detailed info code to explain the sybil score, and would post the panorama of detected sybil behaviors (starlike/chainlike funding network, bulk operation, similar behavior sequence and historical blacklist).

Team Trusta team is top-tier in the AI and security area. The founders are ex-leaders of Ant Financial (the biggest Fintech company in the world with over 1 billion users)AI and Security Labs and are very experienced with ID verification and fraud prevention with proven records in Ant Financial.

Roadmap In the next quarters, Trusta plans to: Product improvement: expand the chain coverage (more layer2 and more non-EVM chains) of TrustScan. Standardize API output for business users. More marketing education regarding the damage of sybils to the communities.