Prior to [[ZKP/Data Marketplace/Tokenized Datasets/Data Ingestion Process|ingestion,]] [[ZKP/Data Marketplace/Tokenized Datasets/Comprehensive Mechanisms of Tokenized Datasets|datasets]] are encrypted using AES-256, a symmetric encryption algorithm with a 256-bit key, widely regarded for its security and efficiency [90]. With a keyspace of 2^256 possible combinations, AES-256 is resistant to brute-force attacks, even with significant computational advancements. It also withstands sophisticated cryptographic attacks, such as differential cryptanalysis, due to its robust substitution-permutation network. This ensures that [[ZKP/Data Marketplace/Security and Privacy/Threat Model and Key Protections|data remains confidential during storage]] on [[ZKP/Data Marketplace/High-Level Overview/Off-Chain Storage with IPFS|IPFS]] and transit across the network, protecting sensitive information like medical records or proprietary models.
The encryption implementation uses the Galois/Counter Mode (GCM) of operation, which provides both confidentiality and authentication, protecting data from both disclosure and tampering. This mode offers several advantages for the [[ZKP/Data Marketplace/Intro|Data Marketplace]] context:
- Authenticated encryption ensures that only parties with the correct key can modify the data, preventing unauthorized alterations during storage or transit.
- Parallelizable design enables efficient encryption and decryption of large datasets, [[ZKP/Data Marketplace/Federated Learning/Federated Learning within the Data Marketplace|crucial for AI applications]] that may involve gigabytes or terabytes of data.
- Minimal expansion of ciphertext compared to plaintext (only by the authentication tag size) maintains storage efficiency, important when dealing with large-scale datasets.
Ownership verification leverages [[ZKP/Data Marketplace/Technical Basis/Cryptographic Foundations/Lifecycle of zk-SNARKs|zk-SNARKs]] through the verification infrastructure, allowing providers to prove they possess the encryption key without disclosing it [77]. In a typical scenario, a provider generates a zk-SNARK proof demonstrating knowledge of the AES-256 key for a dataset, which is then verified on-chain without revealing the key itself. As specified in the base layer, proof generation for a standard 10,000-gate circuit requires approximately 10 seconds, with [[ZKP/ZKP Base Layer/ZKP Blockchain/Technical Build Application Layer/Privacy-Preserving Computations with ZK Wrappers/Circuit Definition for Diverse AI Tasks/Example of PoI Task Circuit for Matrix Multiplication|on-chain verification costing 200,000 weight equivalent]]. This zero-knowledge approach ensures privacy in a trustless environment, a critical feature for the marketplace.
See also: [[ZKP/Data Marketplace/Tokenized Datasets/Metadata Structuring|Metadata Structuring]]