zkAPI: private usage credits for any API (blog.ethereum.org)

The Open Anonymity Project, in collaboration with the Ethereum Foundation, has launched zkAPI, an innovative platform allowing users to pay for metered API usage while preserving their privacy. By utilizing zero-knowledge proofs, zkAPI lets users deposit credits (e.g., ETH, USDC) into an Ethereum vault and authorize API requests without linking their identity to payment. This protects sensitive information in prompts—often related to personal matters like health or finance—from being tied to user profiles or transactions, which has been a significant concern in the current model that requires identifiable API keys. For the AI/ML community, this development is meaningful as it directly addresses privacy issues inherent in utilizing AI services. With zkAPI, even though the AI provider can see the content of requests, they will not know the identity of the person paying for them. The technical framework employs cryptographic techniques, including Groth16 for proofs and Merkle trees to secure deposits, ensuring a high level of transactional privacy. Although there are some limitations regarding network anonymity and potential content leaks due to prompt details, zkAPI provides a groundbreaking approach to maintaining privacy in API interactions, making it a significant step forward for data security in AI applications.