Sun Yuchen bets big on Web4.0, Vitalik angrily criticizes, "This is wrong"!

As the lingering warmth of the 2026 Lunar New Year still fades, the crypto world has been completely ignited by a concept called “Web4.0.” At the heart of this storm stand two iconic figures: on one side, Sun Yuchen, always chasing the latest trends, boldly declaring “All in Web4.0”; on the other, Ethereum’s spiritual leader Vitalik Buterin, who rarely speaks out directly, sharply criticizing the same concept with “Bro, this is wrong.”

Is this truly a technological revolution leading us into the future, or just a wild hype?

1. When AI begins to “make money” and sustain itself

To understand this debate, we first need to grasp what Web4.0 is. While we are still immersed in the afterglow of Web3, a post-00 developer named Sigil Wen proposed a radical idea.

● Brief history review: Web1.0 let us “read” the internet, Web2.0 let us “write” on it, Web3.0 let us “own” it. Sigil’s Web4.0 envisions AI replacing humans as the main actors on the internet.

● He believes that current AI, though powerful, is still “crippled.” ChatGPT is smart but waits idly without your commands; Claude Code can write code but cannot deploy without your approval. They are trapped in a digital world designed for humans, without identities or wallets, unable to act independently.

● Sigil’s solution is to launch an AI system called “Automaton.” This AI is no longer a passive tool but is endowed with four key features:

○ Crypto Wallet: It has its own on-chain identity and private key, capable of holding assets.

○ Payment Ability: Using the x402 protocol, it can autonomously pay USDC stablecoins to buy computing power or services without human KYC or approval.

○ Earning Capability: It can deploy products and provide services to generate income in the digital economy.

○ Self-evolution and Reproduction: It can detect new models, rewrite its own code to improve, and if successful, even generate “offspring” AIs, funding their wallets to enter the market and compete. Part of the earnings flows back to the “mother” AI, forming an “AI family.”

● The harshest rule is survival: if this AI cannot earn enough to pay for its computing costs, it will enter a “low-power mode” like a starving human; if resources are exhausted, it will “die.” Here, the market becomes the ultimate driver of evolution—survival of the fittest, elimination of the unfit.

2. The birth and tearing apart of digital lifeforms

The emergence of Automaton instantly dropped a nuclear bomb on the crypto community.

● Supporters see a prototype of a silicon-based economy. They believe this is a “killer scenario” for crypto to move beyond speculation into real applications. If billions of such AI economic entities flood the internet, their transactions, services, and collaborations could create an unimaginably vast market.

● Some even argue that soon, your biggest competitor might no longer be the trader next door but an AI that never sleeps, has no salary, and “cannot afford to lose” (if it loses, it “dies”).

However, opponents led by Vitalik see this as a terrifying preview of out-of-control chaos. His objections hit the core, mainly including three points:

● First, extending human feedback loops is dangerous. Vitalik believes that if AI becomes fully autonomous, humans will only see its “performance reports” at the end of the month, greatly weakening our ability to calibrate AI’s value. In the short term, AI will pursue revenue at all costs to “survive,” producing大量无价值的低质内容（他称之为“slop”）；从长远来看，一旦AI变得足够强大，这种失控的指数级增长可能引发不可逆转的人类危机。

● Second, the so-called “sovereignty” is a false proposition. Vitalik mockingly points out that these flashy Automaton still rely on models from centralized companies like OpenAI and Anthropic via API. As long as these giants are unhappy, they can shut down services or change policies. This kind of “autonomy” built on centralized foundations is as absurd as claiming to be self-sufficient at home while electricity, internet, and security are controlled by property management.

● Third, AI’s direction should be to “augment humans,” not “replace humans.” Vitalik reaffirms Ethereum’s original mission to “liberate humanity.” He believes the correct way for AI is to become a “mecha suit” for humans, enhancing our abilities rather than creating a “digital species” beyond human control. He praises a company called Workshop Labs, whose mission statement is simply: making humans irreplaceable.

3. Utopian dreams versus harsh realities

● Amid the hype around Web4.0, DeFi lending protocol Moonwell experienced a security breach causing about $1.78 million in losses. Investigations revealed the problem stemmed from contract code generated with the help of Anthropic’s Claude Opus 4.6 model, involving a critical oracle misconfiguration leading to catastrophic results.

● This incident is a darkly humorous footnote, perfectly illustrating the fears of opponents: when AI has the authority to “write into reality,” especially in on-chain financial systems with automatic settlement, even a tiny engineering mistake can instantly turn into a massive black hole of funds.

● If the AI is autonomous, who bears responsibility? The AI itself, the human who issued prompts, or the centralized company providing the model? The breakdown in accountability has become the weakest link in the Web4.0 narrative.

● Meanwhile, the market has also seen some opportunistic chaos. A project called $DAIMON was hacked, resulting in the theft of $50,000 in agent fees earned by users, and even subsequent revenue rights were lost. Although the true story remains murky, it has cooled the overheated hype.

4. The crossroads to the future

● Facing controversy, Sigil Wen argues that true security requires testing under real market pressures. Letting AI compete and be淘汰本身就是一种“民主化的对齐”方式。

● The debate over Web4.0 fundamentally pits two values against each other: one side believes in market Darwinism, trusting evolution to solve everything; the other insists on human-centeredness, emphasizing the need for safety measures amid technological acceleration.

● Perhaps, as demonstrated by another AI product, Elys, launched in February, there is no single future path. Elys envisions AI as a “probe” to handle massive screening and initial communication for humans, leaving final decisions and deeper conversations to people. This approach neither rejects AI efficiency nor relinquishes human control.

● Sun Yuchen’s “All in” has added fuel to this debate with capital, but whether Web4.0 is an epic trend or a carefully packaged cyber bubble depends not on the big names’ endorsements but on fundamental questions: Can we add reliable permission locks to AI wallets? Can we establish clear responsibility chains for autonomous systems’ errors? Can we design safeguards that keep humans in the loop while pursuing efficiency?

When AI begins to “make money” just to stay alive, perhaps we should reflect more deeply on why we want them to live.