The Zero Question: What Anthropic's Stock-Bottom Interview Really Asks About Our Industry's Soul

At the heart of every technological revolution lies an unspoken contract between the builders and the systems they construct. We speak of decentralization, of trustlessness, of code as law, but we rarely examine the quieter infrastructure: the organizational rituals that determine who gets to build the future. Consider the recent report from Beating AI news, a small detail that carries the weight of a tectonic shift. During Anthropic's cultural interview, candidates are now asked a question that cuts through the polite veneer of corporate recruiting: Would you still support the company if the stock went to zero? One candidate, honest to a fault, responded that they would not be happy, and the interviewer's expression reportedly turned to disapproval. This is not merely a hiring practice; it is a confession. A confession that the most influential AI company in the safety-first camp is struggling with a fundamental tension between its stated mission and the material reality of its existence. The question is not about the candidate. The question is about us, the entire infrastructure of the AI and crypto ecosystem, and what we are willing to build on top of an economic base we do not fully trust. To understand this moment, we must place it within the broader context of what Anthropic represents. Founded by defectors from OpenAI, the company has built its identity on a deliberate philosophical split: safety over capability, constitutional AI over raw scaling. Their technical contributions, from the Responsible Scaling Policy to their investment in interpretability research, signal an institutional commitment to ensuring that powerful models do not outpace our ability to control them. This is the "safety-first AI" brand, and it is a powerful one. The market has rewarded this positioning, with Anthropic's valuation reaching astronomical heights, reportedly surpassing $600 billion after a massive funding round in 2024. But this valuation is not merely a financial number; it is a statement of faith. It says that the market believes a company whose explicit goal is to be safe, to be cautious, will also be profitable. The interview question is an attempt to codify this faith at the employee level, to ensure that every individual who joins the ship is willing to stay at the helm even if the storm sinks the vessel. On the surface, this appears noble, a form of principled guardianship that many in the open-source and decentralization community would recognize. After all, we have long argued that true believers build for the long-term health of the protocol, not for the short-term price of its token. But a deeper, more troubling pattern emerges when we examine the report's other data points. The compensation package is over $250,000, a competitive signal. The CEO, Dario Amodei, has expressed concern about money-motivated employees. And candidates are spending thousands of dollars, reportedly $4,600, on coaching to pass this culture fit test. The question is not a noble test of commitment; it is a paradox wrapped in a contradiction, a defensive mechanism against the very commercial engine that keeps the lights on. The core insight here is that Anthropic is undergoing a painful transition from a research organization to a commercial product company, and the interview question is a symptom of a cultural panic. They are facing a structural problem: the pursuit of alignment requires the alignment of capital, but the pursuit of capital often misaligns with the purity of the mission. To attract top-tier researchers, they must pay top-tier salaries. But to maintain their brand, they must filter out those who are perceived as only being motivated by those salaries. The question itself is a pressure test. It is designed to filter for candidates who possess a high tolerance for financial risk, a deep ideological commitment, and a willingness to accept a social contract that prioritizes mission over personal wealth accumulation. In the language of my own domain, this is the difference between a HODLer and a mercenary. The HODLer believes in the future of the network even when the chart looks like a falling knife. The mercenary is there for the exit. The question is an attempt to turn every employee into a HODLer, not of Bitcoin, but of the safety-first thesis. Yet, the very need to ask this question betrays a profound lack of trust. If the belief were truly organic, why would it need to be tested so explicitly? The question is an indicator of internal anxiety, a sign that the current employee base may already be experiencing value drift, and that the leadership is trying to inoculate the organization against the future. From a technical perspective, my experience auditing complex decentralized systems has taught me that the most dangerous flaws are often hidden not in the code itself, but in the assumptions of the developers. I spent hundreds of hours manually auditing Aave V2's interest rate models during the DeFi summer of 2020. I found three critical logic errors that could have led to a $4 million exploit. The errors were not caused by complex mathematical errors; they were caused by a misalignment between the mathematical model and the intended social contract. The code was mathematically perfect but ethically blind. Anthropic's interview question is a form of social contract verification. It is an attempt to ensure that the "code" of their organization, the culture, the decision-making processes, is aligned with the stated ethical framework. However, I see a dangerous flaw in their methodology. By predetermining the "correct" answer, they are not encouraging authenticity; they are encouraging performance. The candidate who wants the job will learn to say that they will support the company even if the stock goes to zero. They will learn to perform mission alignment. This is not building a resilient system; it is building a brittle one. It is a trap that creates a false positive, selecting for actors who are excellent at hiding their true beliefs, which is precisely the opposite of the transparency and trust they claim to seek. The ethical principle of decentralization is not about forcing everyone to have the same values, but about ensuring that a diversity of values can coexist without harming the network. Anthropic is drifting toward a centralized ideological monoculture, which, in my view, is a technical risk. This leads us to the contrarian angle that is often overlooked in the mainstream discourse. The conventional wisdom is that Anthropic's "mission-first" culture is a beautiful thing, a bulwark against the nihilistic profit-seeking of other AI labs. But is that the whole story? Let's apply the pragmatism test. The company's commercial strategy relies on the widespread adoption of the Claude API and enterprise services. To fund the expensive safety research, they must generate revenue. This creates an undeniable paradox: they must prioritize commercial success to be safe. When they ask candidates if they would support the company with zero stock, they are implicitly denying this paradox. They are asking employees to sign a social contract that ignores the material reality of the company's existence. This can lead to a dangerous psychological disconnect. It forces employees to suppress their rational economic concerns, to subscribe to a form of "thought, correct" culture that is not unlike the unhealthy dynamics I've seen in over-centralized DAOs. In those DAOs, the mission was the token price. In this case, the mission is the safety itself. But the effect is the same: a chilling effect on open dialogue. When a candidate honestly says they would not be happy if their compensation goes to zero, and that honesty is met with disapproval, it sends a signal that the company prefers blind loyalty over critical thought. This is a sign of insecurity, not confidence. In the world of cryptography, we call this a "centralization risk." The process of a single point of failure. In this case, the single point of failure is not a server but the very culture that is supposed to be the safety net. This brings us to the second contrarian point, which concerns the industry-wide impact. If Anthropic becomes a filter for "safety-pure" talent, it will create a brain drain in a specific direction. The "safety" faction will go to Anthropic, while the "pragmatist" faction will go to OpenAI or other commercial labs. This creates a dangerous polarization within the AI ecosystem. On the one hand, you have a lab that is obsessed with alignment and safety, but perhaps lacking in the practical, rapid-paced engineering to deploy. On the other hand, you have labs that are extremely capable but may be indifferent to the long-term consequences of their work. This is a false binary. The best system, in any domain, requires both. It requires the dreamer who can see the future and the pragmatist who can build the path to get there. By segregating these types, Anthropic may be doing a disservice to the industry. They are creating a "safety ghetto" where safety is considered a separate, pure concern, rather than an integral part of every technical decision. In my work with the "Soulbound Truths" NFT project, I curated a community of artists who rejected speculation. But I did not reject the pragmatists entirely. I created a credential system that allowed them to exist. We had a diversity of intention. Anthropic's question, as reported, seems to reject that diversity in a binary way. They are asking for a yes or no answer, but the real world is a spectrum. What is the information gain here? The key insight is that the "stock to zero" question is not a test of loyalty, but a test of the candidate's perception of the company's own longevity. When a company asks this question, it is inadvertently admitting that its equity is not a reliable store of value. It is saying, "We know our compensation might be worthless, but we are asking you to accept this risk because you believe in the mission." This is a risky admission. In the crypto market, when a protocol team starts to talk about the "long-term vision" instead of the "technical merits," it is often a sign that the token price is falling. Similarly, this question may be a leading indicator of a compensation problem. In a high-pressure, high-cost environment, if the equity becomes a zero, the employees will leave. The question does not prevent this; it merely delays the inevitable, and it creates a culture of shame around financial concerns. It will make employees less likely to raise red flags about the business model, which is the opposite of the "resilience" they are trying to build. **The most significant contradiction lies in the fact that this interview question is a form of a zero-knowledge proof of a bad kind. It claims to verify a candidate's values without revealing the details of the company's own financial health. It asks the candidate to reveal their risk tolerance, but it does not provide transparency about the company's runway, burn rate, or token vesting schedule. This is an asymmetrical information problem. The company is asking for a leap of faith while holding all the cards. In the open-source world, we demand a high standard of transparency. The code is open, the audits are public. But here, the "code" is the culture, and it is opaque. The candidate is being asked to sign a blank check of loyalty, but the terms of the contract are not fully disclosed. I would argue that this is a form of organizational manipulation. It leverages the candidate's desire to work on meaningful AI safety to extract a concession that they might not make if they had the full picture. It is a subtle power dynamic that is deeply uncomfortable. This leads me to the structural issue of the modern tech "cult" and the systemic risk that this represents. The interview question is a sign of a broader trend where mission is weaponized as a form of compensation. Instead of offering the stock options as a reward for success, the company is asking you to accept that they might be worthless as a precondition for employment. This is not a "mission-driven" culture; it is a "mission-subscription" culture. You pay a monthly subscription fee with your labor and your financial security, and in return, you get the identity of being a builder of "safe AI." This is a dangerous deal because it devalues the human capital. The core of a successful organization, whether it's a decentralized protocol or an AI lab, is not the mission; it's the people. And the people are more than just their commitment. They have a complex set of needs, motivations, and fears. By denying the financial reality of the employees, Anthropic is not protecting them, they are erasing them. And an organization that erases the individual humanity of its employees will eventually lose its way. The industry is watching this with a mix of curiosity and alarm. Other companies are likely to copy this tactic, seeing it as a way to filter out the "less committed" candidates. This would be a catastrophic trend. It would create a race to the bottom of employee financial security. The only way to become a top-tier AI company would be to demand that employees accept the possibility of absolute zero. This is not a healthy equilibrium. It is a race to the bottom of the labor market. It would drive away the very people who have the most to offer: the ones who are talented enough to be able to choose a company that treats them fairly. Instead, the system will be left with those who are desperate or naive, which is a terrible filter for talent. So, what is the way forward? I do not believe that the mission should be abandoned. I believe that the mission should be aligned with the compensation structure. The goal should not be to ask the candidate if they will accept the zero. The goal should be to create a compensation structure that is so aligned with the mission that it never becomes zero. This could be done through a different kind of incentive. Instead of just giving stock, the company could give a "safety stock" that is linked to the achievement of safety milestones. For example, the stock could be paid out if the company successfully implements a particular safety feature or passes a third-party safety audit. This aligns the financial interest of the employee with the mission of the company, not by suppressing the financial interest, but by rewarding it. This is the essence of good protocol design. You create the incentives so that the economic and the social contract are the same. In the crypto world, we call this "game-theoretic alignment." The problem with Anthropic's current approach is that it relies on a "reputation game," where employees are expected to behave in a certain way out of social pressure. But a reputation game is fragile and can lead to a "face" game, where everyone pretends to be something they are not. A well-designed game theory approach uses explicit economic incentives to ensure that the desired behavior is the most rational behavior. This is the opposite of what they are doing. Let me give you a concrete example of this principle from the DeFi audit world. When I identified the flaws in the Aave V2 code, I did not ask the developers to sign a contract saying they would be loyal to the protocol if the token went to zero. I proposed a specific fix that would align the interests of the borrowers and the lenders. The goal was to make the protocol so robust that the token would not go to zero. Similarly, Anthropic's best defense against "mission drift" is not to ask people to accept zero, but to build a business model that is so solid, and a product so safe and useful, that the stock becomes worth more and more. The mission and the market should be the same thing. The question is not whether the candidate will accept the zero; the question is whether the company can build something that is so good it never has to ask that question. In my article "Code as Law, but People as Gods," I wrote about the need for systems that can be resilient during moral decay. But a system is not resilient because it forces its participants to ignore reality. It is resilient because it allows them to face reality and adapt. The Anthropic interview is a sign of a system that is trying to ignore the reality of financial risk, which is a dangerous form of self-delusion. The question "stock goes to zero" also speaks to the heart of the token economy and the fundamental difference between a "safety-first" company and a "token-first" company. In the crypto world, we have a saying: "Don't trust, verify." This is a philosophy that encourages us to not rely on the goodwill of a counterparty, but to rely on the code and the incentives. Anthropic's question is a violation of this principle. It is asking for trust without verification. It asks the candidate to trust that the company's mission is pure, without giving them the tools to verify the company's financial health. It is a one-sided "trust me" statement. In the decentralized world, we believe that "code is law, but ethics is soul." The code is the law, but the ethics is the soul. Anthropic's culture is the "code" of their organization, but their financial model is the "law." The question is an attempt to separate the two. It asks for the soul, but it does not address the law. This is a deeply flawed framework. Now, let's consider the broader consequences for the AI safety movement. This interview is likely to become a topic of discussion in the community. Some will see it as a sign of strength, a willingness to ask the hard questions. Others will see it as a sign of weakness, a desperation to control the narrative. This is a distraction. The real debate about AI safety should be focused on the technical problems, such as how to ensure a model's alignment with human values, not on the moral purity of its employees. The focus on the latter is a sign of a lack of focus on the former. If the company is spending its energy on creating an ideologically pure workforce, it is not spending that energy on creating better safety algorithms. The distraction is a security risk. The biggest vulnerability in any complex system is not the code, but the people who are forced to pretend to be what they are not. It is a form of corruption. As I reflect on my own journey, from translating the Ethereum whitepaper into Portuguese to building ethical frameworks for decentralized governance, I realize that the most resilient systems are not those with the most faith, but those with the most information. The Ethereum whitepaper was not a document of faith, it was a document of engineering. It was a detailed plan of how to build a decentralized world computer. It did not ask the reader to accept that the ETH would go to zero; it showed them how the economic security of the system was designed to prevent that. Anthropic needs to follow this example. Instead of asking their employees to accept the possibility of zero, they need to build a model that makes the zero a statistical impossibility. This is a much harder task, but it is the only sustainable one. And so we arrive at the core of the problem, a problem that is not just Anthropic's, but ours as an industry. The interview question is a mirror of the "fake" of the entire AI and crypto ecosystem. We often talk about the "decentralization" of technology, but we have very centralized our moral values. We have created a culture where a person's worth is judged by their loyalty to a brand, rather than the quality of their work. We have created a market where a high salary and a high mission are seen as mutually exclusive. This is a false dichotomy. The best, most resilient systems, whether they are blockchain protocols or AI labs, are those that can handle the full complexity of human motivation. They are the systems that do not require a person to be a "zero" or a "one." They are the systems that can integrate the "zero" and the "one" into a coherent whole. What is the signal to be watched? The first signal is the reaction of the candidate pool. If the top talent starts to reject Anthropic, the company will be forced to adapt. The second signal is the financial performance of the company. If the stock value is stable and rising, the question will become more and more absurd. If the stock is volatile and falling, the question will become more and more desperate. The third signal is the innovation within the safety space. If Anthropic continues to produce cutting-edge safety research, the mission will be validated by the work, not just the interview. If the research stalls, the interview will be seen as a sign of a lack of substance. I will be watching these signals closely. I will also be watching the broader industry for the adoption of similar practices. If we see other companies adopting the "zero" test, I will know that we are in a new phase of cultural control. That would be a sad day for the industry. But let us not end on a note of despair. Let us end on a note of principle. The decentralized world has given us a language to describe the relationship between the individual and the system. We can use this language to evaluate the practices of any organization, whether it is a DAO or an AI company. The principle is simple: "Don't trust, verify." If a company asks for your trust, ask for the verification. Ask for the details of the financial model. Ask for the transparency of the decision-making process. Ask for the data behind the mission. Do not be afraid to ask the hard questions, because the hard questions are the ones that create a resilient system. The question of the "stock to zero" is a hard question, and I respect Anthropic for asking it. But it is a question that must be asked in a two-way street. If they ask you if you will be loyal, you have the right to ask if they will be transparent. This is the fundamental contract of the open-source world. It is the fundamental contract of a healthy civilization. It is the fundamental contract of a future where AI and humans can coexist. My takeaway is that Anthropic's question is not just about compensation, it is about the definition of value. In the 21st century, we are redefining what it means to create value. The old model was about scarcity: you need to own the land, the oil, the factories. The new model is about abundance: you create value through the network, through the code, through the trust. But this new model of value creation is fragile. It relies on the collective belief of a community. If the community loses its belief, the value is zero. The "stock zero" question is a stark reminder of this fragility. It is a reminder that we are all builders of a cathedral that may never be finished. The question is not if the stock will go to zero, but if we can build a system that is worth more than the sum of its parts. A system that is not a "zero-sum" game, but a "positive-sum" game. This is the challenge of our generation. It is the challenge of the open-source movement. It is the challenge of the AI safety movement. It is the challenge of the decentralized world. We must build a system where the "mission" and the "economy" are not in conflict, but are two sides of the same coin. We must build a system where the people are not forced to choose between their own financial security and the future of humanity. This is the only system that will last. This is the only system that is worth building. In my own work, I have often been asked: "Why do you spend so much time on the ethics of technology?" My answer is always the same: because if we do not, the technology will not be worth building. A code without ethics is a weapon. An AI without a conscience is a bomb. A company without a soul is a zombie. The Anthropic interview is an attempt to create a "soul" for the company. But a soul cannot be created by a test. It is created by the living culture, the open debates, and the willingness to admit when you are wrong. The interview should be a conversation, not an interrogation. It should be a meeting of minds, not a filter of submission. The question should be asked in a way that allows the candidate to be honest, not in a way that forces them to lie. The future of AI depends on the future of the culture that creates it. And the future of the culture depends on the honesty of the conversation. As I sit here in Lisbon, looking at the Atlantic, I am reminded that the decentralized, transparent, and resilient. We must be a community that is not afraid to question the rules. We must be a community that is not afraid to say, "I would not be happy if the stock goes to zero." And we must be a community that is not ashamed of that. Because it is that honesty that will build the trust. And the trust is the only thing that is not zero.