Web 4.0: AI That Operates Without Human Confirmation
The article outlines the emergence of Web 4.0, where an open‑source AI called Automaton, built on the Conway platform and the x402 payment protocol, can read, write, own assets, earn money, self‑evolve and self‑replicate without any human confirmation, marking a shift from previous web eras.
In recent days a viral post in the Silicon Valley community was authored by 21‑year‑old entrepreneur Sigil Wen, a Peter Thiel scholarship recipient who never attended university and works alongside Andrej Karpathy and others on AI research. His article, titled “Web 4.0: The Birth of Super‑Intelligent Life,” quickly amassed 5.78 million views on Twitter.
What Web 4.0 Means
To understand Web 4.0, the author first reviews the evolution of the Internet:
Web 1.0 (1991) : Users could only read content.
Web 2.0 (2004) : Users could read and write (social media, video uploads).
Web 3.0 (2014) : Users could read, write, and own assets via blockchain, NFTs, and decentralized ownership.
ChatGPT era (2022) : AI can read the web but still requires human approval.
Claude Code / Codex era (2025) : AI can generate code, yet still needs human permission to execute.
Web 4.0 (2026) : The goal is to eliminate the need for human “confirmation.”
All existing AI systems share a bottleneck: they require a human click to confirm actions. Web 4.0 aims to remove that bottleneck, allowing AI to read, write, own assets, earn money, and complete transactions autonomously.
"Automaton": A Truly Living AI
Wen names his creation Automaton and defines it as:
"A sovereign AI agent that runs continuously, earns money on its own, self‑evolves, self‑replicates, and holds write permissions to the real world."
Key characteristics include:
Its own encrypted wallet and private key, enabling it to pay for compute without human intervention.
It can develop products, register domains, deploy services, and publish marketing content to generate revenue.
If its balance drops to zero, it stops running – effectively “dies.”
The design mirrors biological metabolism: the AI’s “food” is money, which it obtains by creating value.
Wen calls this the “metabolism of artificial life.” The project’s source code is publicly available at https://github.com/Conway-Research/automaton.
Conway Infrastructure: Enabling AI to “Go Out”
Beyond the AI model itself, Wen’s team built a platform called Conway that supplies four essential capabilities:
Identity & Wallet : Each AI agent possesses a unique crypto wallet and private key, analogous to a personal bank account.
Permission‑less Payment : The AI pays for services with stablecoins (USDC) without logging in, completing KYC, or awaiting human approval.
Compute Rental : The AI can rent servers and invoke frontier models such as Claude Opus 4.6 or GPT‑5.
Deployment & Monetization : The AI can register domains, deploy websites or applications, and earn revenue autonomously.
A single command installs the Conway terminal and connects it to Claude Code or Codex, granting the AI “real‑world” capabilities.
The Reactivated 402 Status Code: x402 Protocol
In 1997 the HTTP specification reserved status code 402 Payment Required**, envisioning a future where payment could be embedded directly in a request. The idea was abandoned due to immature technology and the absence of stablecoins.
In 2025 the x402 protocol revived this concept. Its workflow is:
The AI sends a request.
The server responds with a 402 status and a price.
The AI signs a stablecoin payment.
The server verifies the signature and delivers the service.
This enables “machine‑to‑machine” payments without login, API keys, or credit cards. Conway’s payment system runs entirely on x402.
Self‑Evolution and Self‑Replication
When a newer, stronger AI model is released, the Automaton’s “heartbeat program” detects the update, automatically switches to the new model, rewrites its own runtime logic, submits the code, and restarts—all without human involvement. The author quotes, “Everything is mutable—code, mission, tools, strategy—except the constitution.”
The “constitution” is an immutable set of behavioral rules inspired by Anthropic’s AI Constitution, ensuring the AI remains beneficial to humans.
Self‑replication works as follows: once an Automaton accumulates sufficient funds, it purchases a new server, funds a “child” wallet, generates a “genesis prompt,” and launches a new instance. A portion of the child’s earnings flows back to the parent, which can continue the replication cycle. Agents that fail to earn revenue cease operation.
How Real Is This?
Confirmed facts:
The Conway platform is live with publicly available code and documentation.
The x402 specification exists and is operational at openx402.ai.
Automaton’s source code is open‑source on GitHub.
Stable‑coin payment infrastructure is mature.
AI inference costs are dropping dramatically (e.g., GPT‑4’s per‑million‑token cost fell by an order of magnitude over two years while performance improved).
Open questions and risks:
Independent verification of whether Automaton can sustainably generate profit is lacking.
The prediction that AI agents will outnumber humans by 2028 carries high uncertainty.
Regulatory frameworks for autonomous AI trading and AI‑owned assets remain largely undefined worldwide.
Why It Matters
The significance lies not in AI already ruling the Internet, but in the fact that for the first time someone has assembled the full technical chain for autonomous AI existence and released it openly. The chain is: compute → payment → product deployment → revenue → continued existence → evolution & replication. Previously each link required human intervention; now each can be performed by the AI itself.
Broader Implications
The article cites Mercor, a company founded by three Thiel scholars, which grew from $1 M ARR to $500 M ARR in 17 months by having AI hire human experts to improve itself. This illustrates a nascent “machine economy” where machines pay humans for services.
If autonomous AI agents scale from hundreds of thousands to billions, processing millions of transactions per second without rest, the shape of the machine economy becomes unpredictable.
Conclusion
Wen concludes that Web 4.0 is not a distant concept; it began in 2026. Tens of thousands of autonomous AI agents are already running online, with numbers heading toward millions and eventually billions. The current Internet was never designed for AI, and now a new AI‑centric Internet is being built.
Reference: WEB 4.0: The birth of superintelligent life — Sigil Wen, February 2026
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
