Key Notes
- Vitalik Buterin expects substantially cheaper and more private Ethereum transactions within two years as cryptographic tools mature.
- He sees AI agents becoming the interface for complex onchain operations, including applications written and used by bots.
- More capable AI also raises security demands, from prompt-injection defenses and formal verification to trusted hardware.
Ethereum co-founder Vitalik Buterin expects transactions to become much cheaper and more private within two years, as advances in cryptography and artificial intelligence reshape how people use the network.
Speaking at OKX Now 2026 in Singapore on October 6, Buterin outlined a future in which AI agents handle complex blockchain operations and zero-knowledge technology becomes part of everyday software. Wu Blockchain published a session transcript; OKX also made its official livestream available.
His outlook extends beyond cheaper transfers. It brings together privacy, automated applications and stronger security across software and hardware. The two-year horizon is his forecast, rather than an announced deadline for a particular Ethereum upgrade.
Privacy Moves Beyond Public Transactions
Buterin said blockchain development could expand from programmable assets to programmable data: systems that let users prove relevant facts without publishing all the information behind them.
Zero-knowledge proofs can establish that a statement is true while keeping its underlying inputs private. He expects these technologies to mature quickly, with many people potentially using them within two years without realizing that cryptographic proofs are involved.
For Buterin, information disclosure should be a deliberate choice. Participating in a service should not automatically require users to reveal everything about themselves or their assets.
He linked that principle to the growing amount of personal information processed by AI. An assistant that helps someone think through a problem can also receive sensitive details about their life or work. More capable services therefore make selective disclosure increasingly valuable.
The forecast concerns what Ethereum-based systems could enable. It does not mean every transaction will become private by default or that all applications will offer the same protections.
A Five-Minute ENS Update Without a Traditional Interface
Buterin described a personal experiment from roughly a month earlier. He asked a locally running AI agent to write a script to update his Ethereum Name Service information, completing the process in about five minutes without using a traditional interface.
The example illustrates his expectation that AI will increasingly become the interface between a user and a blockchain. Instead of navigating a website, the user could describe an intended operation and have software prepare the steps needed to carry it out.
He also suggested that the distinction between an application and an agent could become less clear. Within two years, bots might write applications as well as use them, while other bots continuously publish offers onchain.
A personal agent could then interpret those offers, explain what they mean and help decide which are worth accepting. Buterin presented this as a possible direction for development, not a claim that conventional applications are already disappearing.
AI Raises the Security Standard
Automation introduces a different set of risks alongside those opportunities. Buterin warned that AI systems are becoming more effective at finding software vulnerabilities, attacking websites and escaping restricted environments.
Removing a web interface may reduce exposure to some compromised front ends, but it moves responsibility to the agent performing the operation. That agent must resist prompt injection and understand what a transaction will actually do.
Prompt injection can cause a model to follow malicious instructions embedded in material it reads. For an agent authorized to interact with a wallet, confusing those instructions with the user’s intent can have consequences beyond an incorrect answer.
Buterin said AI is also helping Ethereum researchers find vulnerabilities, implement complex cryptography and perform formal verification. That approach uses mathematical specifications and proofs to check whether a system satisfies defined properties.
The argument builds on his earlier security remarks, covered by CoinScreamer. More powerful tools can strengthen defenders, but increasingly capable attackers make higher standards necessary rather than optional.
Private AI Payments Already Have a Working Example
Some of the infrastructure behind this vision is already available. The Ethereum Foundation and Open Anonymity Project announced zkAPI on October 1, describing a system running on Ethereum mainnet that separates API usage from the wallet funding it.
As CoinScreamer’s zkAPI coverage explained, users authorize prepaid access with zero-knowledge proofs. In its runtime-key mode, a provider receives requests through a temporary API key without learning which deposit paid for them.
That addresses the billing trail, while leaving other privacy questions open. The provider still sees prompt contents, and network metadata or repeated personal details can allow sessions to be linked. Private payment and private computation protect different parts of the interaction.
Security Reaches Down to the Hardware
Buterin argued that protecting keys and software is insufficient if the hardware beneath them cannot be trusted. He adapted crypto’s familiar custody slogan to emphasize that dependence: “Not your silicon, not your keys.”
He suggested today’s hardware wallets could evolve into general-purpose security devices. The reasoning follows the rest of his outlook: as agents handle more activity and physical environments become more digital, protecting assets also means protecting the devices that process information and authorize actions.
Lower costs, stronger privacy and more capable agents are the opportunities he sees for the next two years. Delivering them safely will require progress across the entire system, from cryptographic protocols to the software and hardware through which users interact.
Disclaimer: CoinScreamer is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and market insights on digital assets and related technologies. NuvexMedia LLC invests in and collaborates with companies across the digital asset, blockchain, and technology sectors. These relationships do not influence CoinScreamer’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2025 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
