Altman calls for new internet protocol built for AI agents

OpenAI CEO Sam Altman posted on X on Friday that it is time to "seriously rethink how operating systems and user interfaces are designed," arguing that the digital infrastructure underpinning the modern internet was built exclusively for humans and must now accommodate AI agents as first-class participants.

Altman floated the idea of a new internet protocol that works for both people and agents, envisioning a foundational layer with built-in authentication, payments, and data transfer — what he has previously described as akin to a next-generation HTTP. In a November 2025 conversation at Sequoia's AI Ascent, Altman articulated a similar vision: "I hope something in between those that there is sort of like a new protocol on the level of HTTP for the future of the internet, where things get federated and broken down into much smaller components, and agents are constantly exposing and using different tools and authentication, payment, data transfer".

Published

Apr 27, 2026

A Growing Ecosystem of Agent Protocols

The post arrives as a sprawl of competing protocols jockey to become the plumbing of the agentic web. Anthropic's Model Context Protocol handles connections between AI tools and applications, Google's Agent-to-Agent protocol governs inter-agent communication, and newer standards like NLWeb and AGENTS.md aim to make web content machine-readable. Meanwhile, Coinbase and Cloudflare have developed the x402 micropayment protocol, a blockchain-based open standard allowing automated programs to transact directly without human intervention at each step.

World ID and the Trust Layer

Altman's vision dovetails with his separate World project, formerly known as Worldcoin, which unveiled World ID 4.0 at a San Francisco event on April 17. The upgrade introduced what the company calls "full-stack proof of human" infrastructure, with integration partners including Zoom, Tinder, DocuSign, Shopify, and Okta.

Central to the upgrade is AgentKit, first launched in March 2026, which lets verified humans delegate their cryptographic proof-of-human credential to AI agents. Platforms receiving requests from such agents can confirm a real person stands behind the activity without learning who that person is, using zero-knowledge proofs. TechCrunch reported that agent delegation allows a World ID to be tied to a specific agent so that websites know "a verified person is behind the behavior".

Hardware Ambitions Follow the Software Vision

Altman's post landed just hours before supply chain analyst Ming-Chi Kuo reported that OpenAI is co-developing a custom smartphone processor with MediaTek and Qualcomm, with mass production targeted for 2028. Kuo noted that AI agents are "quietly redefining what a phone is" and that OpenAI would need control over both the operating system and hardware to integrate agents directly into a device's real-time context.

The convergence of protocol-level infrastructure, biometric identity, and purpose-built hardware sketches an outline of what OpenAI appears to be building: not just the models behind AI agents, but the entire stack — from silicon to identity layer — through which those agents interact with the world.