What is ERC-7857? A New Standard For Intelligent NFTs

Limitations of Existing NFT Standards

Current NFT standards, such as ERC-721 and ERC-1155, have been effective in representing static digital assets like artwork and in-game items. However, they have significant limitations when applied to AI agents. Here are the primary issues:

1. 1. Static Metadata: ERC-721 and ERC-1155 tokens rely on static metadata that doesn’t change once the token is minted. However, AI agents are dynamic, constantly evolving based on new data and training. The static nature of existing metadata standards makes them unsuitable for representing the continuous learning and adaptation that defines AI.
   2. Insecure Metadata Transfer: When an NFT is transferred, typically only the ownership of the token is transferred, whilst the metadata remains either static or is inaccessible. For AI agents, the metadata—such as neural models, learning history, and behavioral patterns—needs to be transferred along with the ownership to ensure that the new owner has full access to the agent’s capabilities.
   3. No Encryption for Sensitive Data: AI agents often contain proprietary data or sensitive information, such as custom-trained models or user-specific data. Existing NFT standards don’t offer native support for encrypting or securely handling this data, making them ill-suited for tokenising AI agents.

How will ERC-7857 solve these issues?

ERC-7857 was specifically developed to address the shortcomings of current NFT standards, offering a more secure and flexible framework for AI agents. Here’s how ERC-7857 solves the issues:

1. 
   
   1. Privacy-Preserving Metadata: ERC-7857 allows for the encryption of the metadata associated with AI agents. This ensures that sensitive data, such as neural network models or proprietary learning algorithms, remains private and secure.
   2. Secure Metadata Transfer: When an ERC-7857 token is transferred, the associated metadata is also transferred securely. Both the ownership and the metadata are handled together, ensuring that the new owner receives the AI agent in its entirety, including its dynamic capabilities.
   3. Dynamic Metadata: Unlike static metadata in traditional NFTs, ERC-7857 supports the dynamic nature of AI agent metadata. As AI agents learn and evolve over time, their metadata can be updated to reflect changes in their behavior, intelligence, or other capabilities. This ensures that the agent’s capabilities remain intact and are accurately represented.
   4. Integration with Decentralized Storage: ERC-7857 works in tandem with decentralized storage solutions like 0G Storage, ensuring that encrypted metadata is stored securely and remains tamper-proof, whilst still being accessible to authorized parties.
   5. Verifiable Ownership and Transfers: ERC-7857 uses cryptographic proofs and oracles to verify both ownership and the integrity of the metadata transfer. This adds an additional layer of security and trust to the process, ensuring that the transfer of both ownership and metadata is authentic and verifiable.