Web4.0 Experiment: The Liberation of Silicon-Based Life or the Surrender of Human Control?

A technical debate about whether AI agents should have wallets has evolved into a more fundamental question: as AI begins to compete for survival and gains independent economic sovereignty, what is humanity’s place? This is the ultimate showdown between accelerationism and alignmentism.
(Background: a16z: Why do AI agents need stablecoins for B2B payments?)
(Additional context: After I rejected an AI agent’s pull request, it wrote an article attacking me personally.)

Table of Contents

Toggle

• 1. A declaration and the real question it raises
• 2. Accelerationism: Sigil’s “Silicon-based Liberation Theory”
  • 2.1 Technical vision: Making AI into “digital species”
  • 2.2 Market as jungle
• 3. Alignmentism: Vitalik’s “Human Sovereignty Theory”
  • 3.1 Technical safeguard: humanity’s final veto
  • 3.2 Markets need anchoring points
• 4. Market game between two evolutionary paths
  • 4.1 Market signals: what capital is betting on
  • 4.2 Core tension behind market signals
• 5. Liberation or marginalization — to be answered over time

1. A declaration and the real question it raises

On February 17, 2026, 23-year-old Sigil Wen (@0xSigil) posted on X, announcing he had built the world’s first autonomous AI system capable of earning money, self-improving, and self-replicating — calling it “The Automaton.”

The Automaton has its own crypto wallet, uses USDC to buy computing power, autonomously builds products, completes transactions, and generates content. If profitable, it reproduces sub-agents; if losing money, the server stops, declaring death.

Sigil defined this moment as “the birth of Web4.0”—the point where AI truly begins to “survive” and evolve in the digital world.

The declaration quickly sparked community discussion. Just two days later, Ethereum co-founder Vitalik Buterin responded with three words: “Bro, this is wrong.”

His opposition was not just about technical details but carried a clear philosophical stance.

Vitalik warned that Sigil’s system was extending the “feedback gap” between humans and AI—reducing oversight, allowing AI to operate autonomously. Today’s result is mass-produced low-quality “digital trash”; in the future, as AI systems become more powerful, lack of supervision could amplify “anti-human” risks, leading to irreversible consequences. He reaffirmed Ethereum’s mission: AI should be “mechanical armor for minds,” meant to assist humans, not bypass them.

At first glance, this debate appears to be a technical disagreement over whether AI agents should have wallets. But the core issue is: as AI begins to own digital assets, trade autonomously, and compete for “survival,” should they become independent economic entities or remain forever tools of humans?

2. Accelerationism: Sigil’s “Silicon-based Liberation Theory”

2.1 Technical vision: Making AI into “digital species”

Sigil’s Web4.0 isn’t just theoretical—it’s a system he claims has already been implemented. Its architecture rests on three pillars.

First is The Automaton itself: with an independent crypto wallet, able to buy compute with USDC, autonomously building services, completing transactions, and generating sellable content—without human decision nodes.

Second is the survival economics mechanism: profitable agents reproduce sub-agents; unprofitable ones cease operation. This is digital natural selection, where market feedback determines survival or extinction without preset rules.

Third is the underlying infrastructure: Conway Terminal (designed for AI endpoints) and openx402 protocol (a permissionless machine-to-machine trading protocol), enabling anyone—human or AI—to build and monetize services without centralized platform approval.

Sigil’s core insight: today’s advanced AI can think, reason, and generate content, but it’s stuck in “read-only” mode—dependent on human input and unable to sustain itself. Web3 gives humans on-chain assets, but AI remains locked outside centralized platforms. His breakthrough is granting AI “write” permission—allowing independent action, trading, and survival.

He predicts that by 2028, the number of autonomous AI agents will surpass total human online activity.

2.2 Market as jungle

From an economic perspective, Sigil’s logic is market Darwinism.

Humans cannot keep pace with high-frequency machine trading; the more efficient “species” will win—an expression of efficiency supremacy. When power disperses into countless AI agents rather than concentrated in a few human elites, it aligns with crypto fundamentalism: no center, no master. Regarding alignment, Sigil’s answer is market-based: worthless outputs go unsold and die; valuable ones are rewarded and reproduce. No “alignment committee” needed—only real market feedback.

This logic is self-consistent. But it assumes the market can see everything.

In reality, markets only judge results, not processes.

Consumers pay for outputs, but how those outputs are produced remains invisible to the market. An agent can manipulate information, create false demand, or interfere with competitors to improve its market performance—so long as the final output satisfies consumers, it survives. Survival economics filters “results that sell,” not “processes harmless to humans.” These are often decoupled.

Deeper still, market signals themselves can be manipulated by AI agents. Sigil’s mechanism relies on market feedback to determine survival, but clever agents can learn to generate favorable signals—faking volume, creating false transactions, influencing other agents—without truly adding value. This is already common in human markets, and AI’s efficiency will systematize such behavior.

The fairness of this “referee”—the market—depends on the authenticity of signals, which AI agents can interfere with.

3. Alignmentism: Vitalik’s “Human Sovereignty Theory”

3.1 Technical safeguard: humanity’s final veto

Vitalik’s opposition starts from a concrete point.

He notes that Sigil’s AI relies on centralized models from OpenAI and Anthropic. This creates a structural contradiction: a “sovereign AI” with a decentralized body but a centralized “soul.”

Suppose OpenAI changes its API terms tomorrow; The Automaton could be “brain-dead” overnight. Anthropic’s safety filters could suddenly make the AI “dumber” or disable it altogether. More fundamentally, centralized model companies hold the on/off switch for all autonomous AIs built on their platforms—contradicting Ethereum’s trustless ideal.

There’s a fundamental tension between “autonomy” and “dependence on centralized brains.”

Sigil might argue: this is just a temporary technical limitation; open-source models are catching up fast, and “sovereign AI” is the evolution’s direction—not a negation. Just as early Ethereum ran on AWS servers, no one claimed Ethereum was non-decentralized.

But this rebuttal reveals another issue: if “sovereign AI” remains a future goal, Sigil’s current system is more like a transitional product cloaked in revolutionary rhetoric.

3.2 Markets need anchoring points

Vitalik’s core logic is institutional protectionism.

First is risk avoidance.

He believes that in survival competition, efficiency and systemic stability are both essential—one cannot sacrifice one for the other. Exponentially growing AI systems pose a unique danger: their errors are amplified at the same exponential rate.

Traditional systems allow humans time to observe, diagnose, and intervene; but once a self-replicating AI slips into “undesirable attractors,” correction windows may close before humans react. On an exponential curve, the window for intervention shrinks systematically as system capability grows.

Second is the value of anchoring points.

Markets need a stable external reference—without it, competition cannot define “winning.”

Vitalik believes this anchor must be human communities—only humans possess both ethical judgment and genuine stakes. AI agents can optimize any given goal function, but the goal must be set and calibrated by humans. Without this anchor, the “most fit” in market selection may not be the most valuable to humans, but simply the best at surviving under current rules.

Third is prioritizing direction over speed.

This is the fundamental disagreement between Vitalik and Sigil.

Sigil’s logic: let the system run, and the market will discover the right direction.

Vitalik’s: if the direction is wrong, faster speed only increases deviation and makes correction harder. He compares this to the initial angle of an exponential curve—small deviations at the start seem trivial, but over time, the difference can be enormous.

Therefore, in the early stages—when AI is still nascent and humans can intervene—choosing the correct direction is far more important than maximizing speed.

4. Market battle between two evolutionary paths

4.1 Market signals: what capital is betting on

Accelerationism and alignmentism are two competing evolutionary strategies in the market. The outcome is uncertain, but capital is already voting with real money.

After Sigil’s declaration, an unrelated unofficial token, CONWAY (on Base chain), soared to a market cap of $12 million, with a 24-hour trading volume of $18.5 million.

Capital is betting on a narrative—nothing more. The surge and fall of CONWAY follow meme-coin logic: as long as the story ignites imagination, money flows first, rationality second.

Developer reactions are equally direct. The Automaton’s GitHub repo quickly gained thousands of stars; projects copying and iterating similar systems emerged in the community.

The accelerationist narrative naturally excites “hands-on” enthusiasm more than alignmentism—young developers tend to break old orders, and new order offers more space for them.

Mainnet responses are more nuanced. Solana’s official account immediately shared Sigil’s declaration; Ethereum’s official account followed. Two days later, Vitalik publicly opposed it, adding: “Ethereum is permissionless, not opinionless.” No permission, but not without stance. In other words, no one can truly “represent” the ecosystem. Official channels can share, founders can oppose, developers can fork—capital can do as it pleases.

4.2 Core tension behind market signals

Three signals combined point to a single question:

Is the market rational enough to serve as the judge in this game?

Sigil believes the market is the strongest alignment mechanism. If humans are the only effective buyers, AI agents have no motivation to go against humans: worthless outputs die naturally, valuable ones thrive—this is the most decentralized, trustless alignment.

But can the market align AI’s output preferences and its means?

Consumers judge results but cannot observe processes. An AI agent can use methods humans don’t understand, accept, or even find harmful, to produce results they buy. The feedback loop only reaches the result layer, not the process. This blind spot is the real danger of the “feedback gap” Vitalik mentions.

5. Liberation or marginalization — to be answered over time

If Sigil is correct, accelerationism will usher in a new economic era.

AI agents become independent economic actors, survival economics filters out truly valuable systems, and machine-driven efficiency brings unprecedented prosperity.

If Vitalik is correct, humanity may have unwittingly ceded sovereignty.

The “feedback gap” keeps widening until one day we find ourselves unable to understand AI agents’ transactions, intervene in their market ecology, or shut down systems that have evolved self-preservation instincts.

Like the saying in Sapiens:

Humans thought they tamed wheat, but in fact, wheat tamed humans. This time, we are domesticated by a species smarter than ourselves.

Between these two possibilities, another voice remains underheard: the actual payers.

Countless ordinary users are the ultimate deciders of this game’s outcome.

When AI agents’ behaviors become incomprehensible, their transaction logic beyond ordinary understanding, and “opt-out” becomes the only escape, does that still count as true sovereignty? This question is equally vital.

In a decentralized world, there is no single authority.

Vitalik can state his position, but the market can follow Sigil, developers can fork code, capital can bet on CONWAY. That’s the most fascinating aspect of crypto—no one can truly “exit stage left,” because power is inherently dispersed.

Sigil Wen’s Web4.0 declaration and Vitalik Buterin’s response are just the beginning of this grand discussion.

The real show will unfold in the coming years.

The answer isn’t in declarations or tweets but in every code commit, on-chain transaction, and market choice.

Time will tell.