We detect fake NFTs using AI
NFT market ballooned to $41B in 2021. NFT enabled digital scarcity to naturally abundant class of assets (digital art), making it attractive to the art collectors. The value of NFT is driven by the uniqueness / rarity of the underlying asset. However, while blockchain tech guarantees uniqueness and immutability of the metadata (i.e. “digital certificate of ownership” in the ledger), it does not validate the content itself is unique and authentic. This enables “NFT clones”, that cash in on successful NFT collections by releasing similar or even identical copies of the art under similar sounding brands. While the crypto community is still debating whether NFT cloning is piracy or it’s a natural part of the open-source culture, it’s safe to assume that lack of transparency around it can damage the trust in NFT art concept all together. The goal of Duplicat is to provide a fast and reliable way to validate uniqueness of NFT content, by finding duplicates and similar prior art across blockchains.

Problems we are solving
Detecting NFT counterfit
Find identical or intentionally modified art, minted multiple times
Detecting cross-chain clones
Find identical art minted on a different blockchain
Finding off-the-chain prior-art
Find similar or identical art in public domain, that has not been minted as NFT yet
Identifying derivative art
Find similar, but not identical images such as remixes, parodies, etc

Who is it for?

Check if the NFT that is being added to the marketplace is unique and not a copy of an existing art.

Why? To avoid copyright claims and de-listings that anger collectors and creators.

Check if the NFT you are interested in is in fact unique or if there are similar assets owned by someone else.

Why? To make sure you are buying the real thing.

Get notified if when a copy of your art or a derivative is minted by someone else.

Why? To avoid your buyers being upset over non-uniqueness of the art they purchased.

Q: Which content types do you support?
We started with images and eventually expand to audio, video and 3D.

Q: What's your business model?
Option 1. Duplicat is a classic “API-tech” – b2b saas with API for marketplaces, with a pricing model based on number of unique users, assets, QPS, etc

Option 2. Duplicat is a decentralized protocol / smart contract that automatically charges a small fee for validating an NFT transaction on purchase

Q: What are the key tech components?
  • The AI - neural net that can reliably and fast detect indentical and similar images, but in the future - video and audio too
  • Crawler - continuously crawls all blockchains that allow NFT (Ether, Solana, Polygon, etc), downloads all the underlying assets.
  • Validator - for a given digital asset runs a CV image retrieval process to validate against all the indexed assets. Returns matches if found and similarity score for each.
  • API - enables Marketplaces, wallets etc to query Validator in a fast and reliable way
  • UI - web UI and Chrome extension that displays NFT check results for Collectors and Creators

Q: What are your quality metrics?
  • Match precision
  • % blockchain covered
  • Crawling refresh rate
  • Request latency
  • Uptime
Andrey Doronichev
Founder, CEO
As a Director of Product at Google lead YouTube Mobile, Daydream and Stadia
Roman Doronin
AI lead
Founded Eora AI lab, producing more than 65 AI projects. The winner or mutiple AI hackathons.
Vlad Vinogradov
Eng lead
CTO of Eora AI lab, x-Intel. 25+ Computer Vision projects. Winner of mutiple AI hackathons
Contact us:
San Francisco,
Made on