ERC2006 (An AI Model Friendly NFT Standard)

ERC2006 is an experimental crypto-asset standard for AI models. There are several issues in the application of mint and trade AI/ML models in the existing NFT standard. First, AI models are much more complex compared to the traditional content of NFT, such as images and videos. Based on this, the existing single NFT mechanism cannot perfectly fit with AI/ML models. Second, some publicly owned data or models are owned by many individuals. However, multi-signed solutions are too clumsy to solve a large number of multi-party authorization. Third, the existing NFT standard could not protect the metadata of the assets in the process of assets circulation, which means any unauthorized individuals can access the metadata of the assets. It will cause security risks in the circulation of data-related assets, especially personaldata.

To deal with these issues and realise the integration of AI model and NFT, DNFT team present the AI data-friendly ERC2006 standard. It offers three more core functionalities compared with the ERC721 standard.

First, it allows individuals to gather multiple NFTs and pack them into one NFT. It is friendly for packing multiple personal data into a large set of data. It is convenient for the creation and transfer of a large set of data. Second, it allows the NFT asset owner to fragmentize their asset to an ERC20 token. It will be a potential solution to deal with the ownership of huge data sets, especially public data. Third, it contains a copyright protection mechanism by using ZKP, MPC and other cryptographic techniques to control authorized viewing.