Ethereum Foundation launches zkAPI for private payments using zero-knowledge proofs

How zkAPI works

zkAPI allows users to deposit ETH, USDC, or other supported crypto assets into a smart contract vault on the Ethereum blockchain. Once deposited, the user’s balance is represented privately, preventing exposure of account details during standard API interactions.

Traditionally, API access relies on static API keys linked to specific user accounts and payment methods, creating a direct link between users and their service usage history. zkAPI changes this by issuing temporary, single-purpose spending credentials, which cannot be traced back to the original deposit.

To access a paid API service, a user’s device generates a zero-knowledge proof confirming they possess adequate funds and preventing double spending. This proof is validated by the zkAPI server, which then provides a short-lived API key with a set spending limit. Requests can be submitted directly to service providers—such as AI applications—without exposing the user’s vault or payment details.

Mini dictionary: Zero-knowledge proof (ZKP): A cryptographic method allowing one party to prove knowledge of specific information to another party without revealing the information itself. In crypto applications, ZKPs help maintain privacy and security during transactions or authentication processes.

This setup ensures that the payment server never views the prompt or content submitted to the AI application, while the AI provider cannot identify the wallet that funded any particular request.

Initial use cases and future potential

The Ethereum Foundation identified AI inference as the primary use case for zkAPI at launch, citing the sensitive nature of AI prompts and the increasing demand for privacy in data-driven applications. However, the Foundation emphasized that zkAPI’s technology is adaptable and not limited to artificial intelligence applications alone.

AI prompts often contain deeply personal or confidential information, making user privacy essential when interacting with such services. zkAPI enables per-request payments without exposing users’ financial activities or requiring long-term account relationships.

The infrastructure supporting zkAPI is also compatible with blockchain RPC endpoints, image and video generation platforms, bandwidth metering for VPN services, and machine-to-machine payments. This would allow service providers to receive payments on a per-request basis, reducing the need for persistent accounts or long-term user histories tied to payments.

Limitations and privacy considerations

Although zkAPI breaks the link between payments and service usage, it does not guarantee absolute anonymity. While the payment relationship is hidden, the AI service provider continues to see the prompt text itself and may receive network-level data such as IP addresses.

Additionally, repeated user patterns or embedded personal information in prompts can still enable linking across different sessions, even if financial details remain private. The Ethereum Foundation has acknowledged these privacy limitations and suggests users should remain aware of potential indirect data exposures.

The privacy model centers on unlinking payments from service requests rather than hiding all identifying information. Users must consider residual risks, particularly when sensitive data is involved.