The Ethereum Foundation and the Open Anonymity Project have launched zkAPI, a zero knowledge payment system that allows users to pay for AI inference and other metered APIs without tying every request to a persistent billing identity. Users deposit credits such as ETH or USDC into an Ethereum vault and later authorize limited API spending with zero knowledge proofs instead of exposing an account, card, or persistent wallet identity.

The launch is particularly notable because EtherWorld covered the original idea in February, when Buterin and Crapis proposed ZK API usage credits as a way to make Ethereum infrastructure more suitable for AI and autonomous agents. What was then primarily a research design is now running on mainnet.

zkAPI Breaks the Link Between Payment & Identity

Most AI APIs rely on a straightforward account model. A user creates an account, adds a payment method and receives an API key. Every request made through that key can then be associated with the same billing identity and usage history.

That architecture becomes more sensitive as AI systems handle increasingly personal or confidential data. Prompts can contain business plans, financial information, private communications, internal research or other material that users may not want connected to a long term billing profile.

zkAPI takes a different approach. A user first deposits credits into an Ethereum vault. Instead of maintaining a conventional account balance tied to an identity, the funds become represented by a private note.

When the user wants to access a paid API, software running locally generates a zero knowledge proof showing that a valid funded note exists and can cover the requested spending amount. The zkAPI server verifies that proof without learning which specific deposit created the balance or who controls it.

The system then creates a short lived API key with a fixed spending limit. Prompts can be sent directly to the AI provider using that temporary key. Once the session ends, usage is recorded through a signed receipt and the appropriate amount is deducted from the private balance.

The important architectural separation is that the payment side knows a legitimate spend occurred, while the API provider knows what service was requested. Neither side is supposed to automatically learn the complete link between the original funding source and the API activity.

That direction fits Ethereum's broader privacy strategy. Earlier this year, Vitalik outlined work toward making privacy more native across Ethereum, including improvements around account abstraction, transaction inclusion and access layer privacy.

How zkAPI Uses Zero Knowledge Proofs

zkAPI's privacy model relies on several cryptographic components working together. Deposits are represented as cryptographic commitments inside a Merkle tree. A user can therefore prove that their private note belongs to the set of legitimate balances without revealing which particular deposit they own.

Every spend also produces a nullifier, a one way identifier derived from the underlying note. If a user tries to spend the same balance twice, the same nullifier appears again, allowing the system to detect the double spending attempt without exposing the user's identity.

The current implementation uses Groth16 zero knowledge proofs on the BN254 curve, Poseidon hashing for commitments and nullifiers, and a Merkle tree with a depth of 32. Spend proofs are verified offchain, while the Ethereum smart contract handles actions including deposits, closing balances and exiting the system.

One important property is that funds do not depend entirely on the continued operation of the zkAPI service. The Ethereum Foundation says users retain an onchain exit mechanism through the vault contract, meaning they can withdraw their funds even if the zkAPI servers disappear.

Ethereum's roadmap increasingly places ZK systems at the center of both scaling and verification. EtherWorld previously explored how Ethereum's longer term roadmap now emphasizes STARKs, privacy, formal verification and post quantum security.

Tim Beiko has also discussed a future where ZK-EVMs and AI increasingly intersect with Ethereum's infrastructure. The Enterprise Ethereum Alliance recently launched a Privacy Working Group focused on technologies including ZK proofs, MPC and TEEs.

Today the EF and OA announce zkAPI—a new means for private AI

Picture this: deposit ether into a vault, sign a zk proof, get a fixed amount of private inference. Advances like these help us move past middlemen's requirements that we link our data and identities.

Read more below https://t.co/LBy0Xz6cyi pic.twitter.com/7tOHRmXM4r

— Ethereum Foundation (@ethereumfndn) October 1, 2026

AI Is Only the First Use Case

According to the Ethereum Foundation, the same system could be used for blockchain RPC queries, image generation, video generation, VPN bandwidth and machine to machine services. This becomes more significant as autonomous AI agents begin interacting with blockchain infrastructure.

Instead of creating traditional user accounts and storing persistent API credentials, an autonomous agent could theoretically hold funds, authorize limited usage and purchase services programmatically. EtherWorld has documented the growing infrastructure landscape in Who's Building for Ethereum's AI Agents?, where identity, payments, verifiable computing and autonomous execution are beginning to converge.

Other proposals are addressing complementary problems. For example, WYRIWE introduces verifiable input provenance for AI agents, attempting to prove that the input a user approved is the same input ultimately sent to an AI model.

The Ethereum Foundation had already supported the Open Anonymity Project before zkAPI's mainnet launch. Its Q2 2026 allocation report described the project as infrastructure intended to enable unlinkable access to gated APIs, beginning with AI inference.

zkAPI Improves Payment Privacy, Not Total Anonymity

An AI provider still needs access to the prompt in order to run the model. The provider can therefore see the content of the requests it processes.

If a user repeatedly accesses the service from the same IP address, sessions could potentially be correlated even if their payment activity remains cryptographically unlinkable. The Ethereum Foundation suggests that users seeking stronger network privacy could combine zkAPI with technologies such as Tor.

Prompt contents themselves may also reveal identity. Repeated personal details, distinctive writing patterns, conversation history or reused project documents can allow separate sessions to be associated with the same person.

Ethereum's privacy direction increasingly acknowledges that reality. Work around native Ethereum privacy spans wallets, transaction propagation, account abstraction and private reads rather than depending on one universal privacy mechanism.

The same trend is visible in institutional Ethereum. Projects such as ZKsync and BitGo's privacy focused banking infrastructure are exploring how public blockchain settlement can coexist with selective confidentiality rather than forcing every piece of data onto a transparent ledger. zkAPI applies that principle to digital services. It does not attempt to hide everything.

To promote your Web3 articles, events, and projects, you may reach out anytime via EtherWorld PR for submissions and collaboration.

Related Articles

• Vitalik’s ZK API Proposal Aims to Make Ethereum the Home for AI
• Vitalik Highlights Ethereum’s Short-Term Push Toward Native Privacy
• Ethereum Foundation Reveals Its New Playbook for Ethereum’s Future
• Ethereum’s Roadmap Just Changed: Here’s What’s Next
• Ethereum’s Next 10 Years: Tim Beiko Outlines Ethereum’s Roadmap

To follow blockchain news, track Ethereum protocol progress, and read our latest stories, subscribe to our weekly today.

Join the EtherWorld & Avarch Internship Program and build your career in blockchain, content, social media, video, podcast editing, or operations. Send your resume and brief introduction to contact@etherworld.co.