Hook
On August 19, The Information reported that Anthropic, the AI safety darling with a $18 billion valuation, is preparing a class of stock with enhanced voting rights for CEO Dario Amodei and other co-founders. This move, designed to shield the founding team from external shareholder pressure, marks the first time Anthropic’s leadership will hold super voting rights. Superficially, this is a governance tweak. But for those who have spent a decade dissecting the architecture of trust, it is a revealing structural flaw: a protocol that claims to prioritize safety yet centralizes control in a single point of failure. The irony is not lost. Anthropic’s mission is to build safe, beneficial AI, yet its governance model mirrors the very systems it seeks to replace: opaque, hierarchical, and resistant to accountability.
Context
Anthropic was founded in 2021 by former OpenAI employees, including Dario and Daniela Amodei, who left citing concerns over OpenAI’s shift toward commercialization. The company positions itself as a “public benefit corporation” and has attracted significant investment from Google, Salesforce, and other institutional players. Its cap table is heavily diluted; the co-founders hold a relatively low percentage of equity. To maintain control, they are now introducing a dual-class share structure, similar to those used by Facebook, Google, and Snap. Concurrently, they plan to retain a non-shareholder trustee system and use special class shares to elect a majority of board members.
This is not a new playbook. In the crypto world, we have seen similar patterns: projects like Uniswap, MakerDAO, and Compound initially promised decentralized governance, only to retain core team control through treasury multisigs, founder vetoes, or token distribution mechanisms that concentrate voting power. The promise of “code is law” often crumbles when the code is written by a committee that can change the rules. Anthropic’s move is a textbook case of what I call “governance theater”: the appearance of democratic oversight while the true power remains in a small, closed circle.
Core
Let me dissect this systematically, using the same forensic lens I applied to the Parity Wallet vulnerability in 2018 and the Terra/Luna collapse in 2022.
1. The Liquidity of Control
Super voting shares are a form of governance liquidity. They allow founders to retain control while raising capital from external investors who expect a return. The problem is that this creates a misalignment of incentives. Shareholders are effectively buying a product with no voting power, while the founding team can make decisions without accountability. In the crypto context, we have seen this lead to catastrophic outcomes: the DAO hack, the Bitfinex-Tether debacle, and the FTX collapse. In each case, a small group of individuals held disproportionate power, and when that power was abused, the system collapsed.
Anthropic’s argument is that this structure protects the company from short-term profit pressures. But as a risk management consultant, I can tell you that protection from external pressure is a double-edged sword. It also insulates the team from market feedback. When a company cannot be easily voted out, it becomes a monarchy. And monarchies, historically, are not known for their safety.
2. The Trustee System: A Black Box
Anthropic’s plan to retain a non-shareholder trustee system is even more concerning. A trustee system is a governance layer that exists outside the shareholder structure. In traditional finance, trustees are used to manage assets on behalf of beneficiaries. In the crypto world, we have seen similar mechanisms in the form of “foundation councils” or “governance committees” that are not elected by token holders. The problem is that trustees are often unaccountable. They are appointed by the founders, and they can be removed only by the founders. This creates a governance vacuum.
I have seen this before. In my 2020 analysis of Compound Finance, I identified that the governance token distribution was massively skewed toward early investors and the team. The protocol’s “decentralized” voting system was a chimera; actual power resided in the hands of a few whale addresses. When Compound proposed a change to its interest rate model, the vote was won by a single address that held 20% of the supply. The result was a protocol that served the interests of the few, not the many.
Anthropic’s trustee system is worse. It is not even a token-weighted vote; it is a closed-door appointment. The trustees are not elected by shareholders or by the public. They are chosen by the founders. This is the antithesis of transparency.
3. The Math of Power Concentration
Let’s quantify the risk. Assume the co-founders hold 10% of the equity (a generous estimate, given dilution). With super voting shares, they can control 50% or more of the voting power. This means that a group representing 10% of the economic interest can dictate the direction of the company. The remaining 90% of shareholders—including employees, early investors, and public market participants—have no real say.
This is a structural fragility. If the co-founders make a bad decision, there is no mechanism to correct it. The company cannot be acquired against the founders’ will. The board cannot be replaced. The only check is the market, but the market cannot vote out a CEO who holds super voting shares.
In the crypto world, we have seen this play out. The Ethereum Foundation, while not a company, has a similar structure: Vitalik Buterin holds significant influence, but he is not a dictator. The difference is that Ethereum has a public chain, a community of developers, and a fork mechanism. Anthropic has no such escape hatch. If the founders decide to pivot to a commercial model that compromises safety, there is no way for the community to fork the company.
4. The Precedent of AI Safety
Anthropic’s mission is to build safe AI. But safety is not achieved by concentrating power. In fact, safety is a property of distributed systems. The more centralized the control, the more catastrophic the failure. This is a lesson from aviation, nuclear power, and, yes, blockchain. The Terra/Luna collapse was a failure of a centralized stability mechanism. The SolarWinds hack was a failure of a centralized software supply chain. The FTX collapse was a failure of a centralized balance sheet.
Anthropic’s super voting shares are a form of centralized control. They reduce the number of eyes that can scrutinize decisions. They reduce the number of voices that can challenge the orthodoxy. They create a single point of failure.
Contrarian
Now, let me play the contrarian. The bulls might argue that super voting shares are necessary to maintain the company’s mission. They point to Larry Page and Sergey Brin at Google, who used dual-class shares to resist short-term profit pressures and invest in long-term projects like self-driving cars. They point to Mark Zuckerberg at Facebook, who used super voting shares to acquire Instagram and WhatsApp despite shareholder opposition. They argue that the founders’ vision is the company’s greatest asset, and that external shareholders would destroy that vision.
There is some truth to this. In the crypto world, we have seen projects sacrifice long-term sustainability for short-term gains. The DeFi summer of 2020 was a race to the bottom, with protocols offering unsustainable yields that collapsed within months. The founders of these projects often had little control, and they were forced to compete for liquidity. In contrast, projects like Bitcoin and Ethereum have resilient governance because they are not controlled by a single entity.
But the contrarian argument misses a key point: the difference between leadership and dictatorship. Leadership is about influence, not control. Vitalik Buterin has influence over Ethereum, but he cannot unilaterally change the protocol. He must convince the community. Satoshi Nakamoto is gone, and Bitcoin thrives. The founders of Anthropic are not Satoshi. They are not even Vitalik. They are former employees of a company that has already lost its way.
Moreover, the track record of dual-class shares is not stellar. A study by the Harvard Law School found that companies with dual-class structures underperform over time. The founders’ vision may be valuable, but it is also fallible. And when it fails, there is no correction.
Takeaway
Anthropic’s super voting shares are a governance failure waiting to happen. They are a structural vulnerability that will be exploited not by hackers, but by the founders themselves, whether through incompetence or hubris. The company’s mission of AI safety is undermined by its own governance model.
Logic survives the crash; emotion dissolves. Precision is the only antidote to chaos. Clarity cuts deeper than noise.
The question is not whether Anthropic will fail, but when. And when it does, the post-mortem will reveal a body that was dead on arrival: a centralized system that claimed to be different but was, in fact, the same as every other corporate empire.
In the crypto world, we have a term for this: exit liquidity. The founders will enrich themselves while the shareholders and the public pay the cost.
The math doesn’t lie. Exit liquidity is not a feature. Audits are opinions, not guarantees. Rationality is scarce. Code compiles. Lies don’t. Volatility reveals character.
Anthropic’s super voting shares are a step backward. They are a betrayal of the principles of transparency, accountability, and decentralization that the crypto community holds dear. But they are also a warning: the same dynamics that corrupted the banking system and the internet are now corrupting AI.
We have been warned. The question is whether we will listen.
