MORGANABLE BUSINESS /MARKETS
The Ethereum Foundation has launched zkAPI, a new payment system designed to give users more privacy when paying for artificial intelligence services and other metered application programming interfaces.
Akure —
The Ethereum Foundation has launched zkAPI, a new payment system designed to give users more privacy when paying for artificial intelligence services and other metered application programming interfaces. The system went live on Ethereum mainnet on October 1, 2026, according to the foundation.
The project was developed with the Open Anonymity Project. It addresses a growing concern around AI services: payment information and usage records can often be connected to the same account. As a result, providers may be able to associate a customer with a long history of API activity.
With zkAPI, the Ethereum Foundation aims to separate payment from identity. Users can deposit credits into an Ethereum vault and later prove that they have enough funds to pay for an API request without revealing which deposit belongs to them.
The system uses zero-knowledge proofs to make that separation possible. In simple terms, a zero-knowledge proof allows software to demonstrate that a statement is true without revealing the private information behind it.
Under traditional systems, an API key is normally connected to an account. That account may also contain billing details and a record of previous requests. Therefore, a provider can potentially build a long-term profile around a customer’s activity.
zkAPI attempts to break that connection. After a user deposits funds, the system creates a private balance that can be used for metered services. The user then generates a proof from their own device when they want to authorize spending.
The payment server checks the proof and can issue a short-lived API key with a spending limit. The key does not permanently identify the user. Instead, it allows a session to operate within a predefined spending cap.
Next, the user’s prompts can go directly from the device to the AI provider. The provider receives the request and returns the response, while the payment system handles the financial settlement separately.
The system also settles actual usage rather than simply charging the full spending limit. When a temporary key expires, the provider records the amount used in a signed receipt. The payment server can then deduct the corresponding amount from the user’s private balance.
Consequently, one authorization can cover an entire session instead of requiring a separate blockchain transaction for every AI request. That approach is intended to make private payments more practical for services that charge according to usage.
The Ethereum Foundation says zkAPI can support more than AI services. Potential applications include blockchain remote procedure calls, image and video generation, virtual private networks, bandwidth services and machine-to-machine payments.
AI inference is the first major use case because AI prompts can contain personal or commercially sensitive information. Users may ask models about finances, private projects or other subjects they do not want connected to their payment history.
However, the new system does not make every part of an AI interaction anonymous. The foundation specifically notes that zkAPI does not hide the contents of prompts from the AI provider.
Prompt content can create another privacy risk. If users repeatedly provide the same personal details, writing patterns or project information, a provider may potentially recognize that the requests come from the same person.
Therefore, zkAPI focuses on separating payment identity from API usage rather than providing complete online anonymity. Users who need stronger network privacy may need additional tools that address network-level identification.
Technically, zkAPI uses several cryptographic components. The system uses Groth16 zero-knowledge proofs on the BN254 curve, while Poseidon is used for hashing commitments and nullifiers. Deposits are organized through a Merkle tree.
The Ethereum Foundation says the vault operates through a smart contract on Ethereum. This means users can close their balances and withdraw their funds onchain rather than relying entirely on a traditional company account.
The project also includes a local client and software development tools. These are designed to work with existing OpenAI-compatible applications and other tools. The foundation says users can point compatible applications to the local client without changing their normal workflow.
In addition, zkAPI offers a proxy mode. In that setup, the zkAPI server relays requests to the service provider. The option may be simpler to operate, although the relay can see network traffic. The runtime-key model provides a stronger separation between payment and service traffic.
The launch follows an earlier research proposal on zero-knowledge API usage credits. Ethereum Foundation dAI lead Davide Crapis and Ethereum co-founder Vitalik Buterin published that design in February 2026. The latest release turns the concept into a working mainnet implementation.
The development also reflects a broader effort to connect blockchain technology with AI infrastructure. As AI agents increasingly interact with paid digital services, developers need ways for software to make payments without necessarily creating a traditional account for every transaction.
For businesses, the system could provide another method of accepting usage-based payments. Providers could verify payment proofs and settle signed usage receipts while keeping their existing pricing and rate-limit structures.
Still, zkAPI is not a universal privacy solution. Its benefits depend on how users connect to services and what information they send. Network metadata and recognizable prompt content can still weaken privacy.
The Ethereum Foundation has made the system available on Ethereum mainnet, where its vault and related software can be used. The project therefore moves zero-knowledge API payments beyond the research stage and into practical testing.
Overall, zkAPI introduces a new approach to paying for AI and other metered digital services. By separating payment authorization from identity, it seeks to reduce the amount of billing information connected to API activity.
Its launch could also provide a foundation for privacy-focused payments for users and developers across digital services.
