The Regulatory Capture Accusation: When AI Safety Becomes a Moat

David Sacks, the venture capitalist and former PayPal COO, has thrown a sharp accusation at Anthropic. He claims the company, creator of the Claude model, is actively pursuing regulatory capture against the open-source AI community. His argument is simple: by pushing for strict safety regulations, Anthropic is building a compliance moat that only well-funded, closed-source players can afford. The open-source ecosystem, with its community-driven development and distributed contributors, would be squeezed out.\n\nThis is not a new accusation, but it lands with unusual force. Sacks has a platform. And he is not a fringe voice. He is a prominent figure in Silicon Valley's libertarian wing, a man who has built his reputation on betting against the status quo. His accusation turns the entire AI safety narrative on its head. Safety becomes a weapon, not a principle.\n\nAnthropic has built its brand on being the "safe" AI company. From its founding principles, it has positioned itself as the counterweight to the more reckless race for scale. Its public messaging emphasizes alignment, red-teaming, and careful deployment. This positioning is not just ethical. It is a market strategy. The company's valuation and its enterprise adoption are heavily dependent on this narrative of trust. When an AI system makes a mistake, the public blames the technology. But when a company has a reputation for caution, the blame is softened. The narrative of safety becomes a shield.\n\nBut is this shield being converted into a sword? Sacks' accusation points to a more insidious dynamic. The AI regulation space is heating up. The EU AI Act is passing. The US is debating its own framework. In this legislative vacuum, the voices of the largest players carry disproportionate weight. They have the lobbyists, the legal teams, and the public relations machinery. When Anthropic speaks at hearings about the dangers of unregulated open-source models, it is making a true observation about risk. But it is also advancing its own commercial interests. The cost of compliance, which includes rigorous testing, documentation, and liability insurance, is a fixed cost. For a company with millions in funding, this is manageable. For a small startup running a fine-tuned Llama, it is a death sentence.\n\nThe open-source community has a counter-narrative. They argue that open models are safer in the long run. They point to the transparency of the code, the public scrutiny, and the ability for anyone to audit the model. Closed models, they argue, are black boxes. The danger is hidden behind a corporate firewall. This argument has merit. But it is also an argument that is easy to make when you have nothing to lose. The open-source community does not have the same legal liabilities. They are not responsible for the deployment of a model by a third party. They do not have a balance sheet to protect. Their risk is theoretical. Anthropic's risk is existential.\n\nLet's look at the underlying incentive structure. My time auditing smart contracts in the 2017 ICO boom gave me a clear lens on this. The pattern is identical. The founders who talked the most about "security" and "compliance" were often the ones building a protective moat against the inevitable bear market. The term "regulatory capture" is not just a political buzzword. It is a quantifiable economic strategy. When a company influences the rules of the game, it creates a competitive advantage that no amount of code or innovation can overcome. The market is no longer a free market. It is a governed market. The question is who writes the rules.\n\nAnthropic's stance on safety is not hypocritical. It is logical. They are building a product that requires trust. The trust must be institutionalized. The regulatory framework is the institutionalization of that trust. If the regulators say that only models with a certain level of safety certification can be deployed, then Anthropic wins. They already have the certification. The open-source community, with its decentralized and chaotic development process, does not. This is the structural logic. It is not a conspiracy. It is a rational alignment of incentives. The accusation of "regulatory capture" is just the name we give to the rational behavior of a rational actor.\n\nBut here is the contrarian angle that the crypto and open-source world often misses. The accusation itself is a double-edged sword. Sacks is using his platform to fight for open-source. But his motive is not purely altruistic. He is a venture capitalist. He has a portfolio to protect. The open-source ecosystem, while decentralized, is a source of speculative investment. The narrative of "open-source is the future" is a valuable narrative to sell to LPs. He is not just fighting for a philosophy. He is fighting for the survival of his own asset class. The crypto world learned this lesson in 2021. When the regulators came after the exchanges, they used the narrative of "investor protection" to justify their actions. The exchanges, in turn, used the narrative of "freedom" to fight back. Both sides were using the same mechanism of narrative control. The only difference was the side of the equation they were on.\n\nThe technical reality is that open-source models are getting better at a staggering pace. Llama 3 and Mistral have proven that the gap between open and closed is shrinking. The performance gap is no longer a moat. The safety and regulatory gap is becoming the new frontier. If the open-source community cannot organize itself to handle the compliance burden, the entire ecosystem will be pushed into the shadows. We have seen this in DeFi. The crypto community, after the 2022 crash, was forced to comply with travel rules and licensing requirements. The cost of compliance drove small players out. The industry consolidated. The same thing is happening in AI.\n\nThe real question is not whether Sacks is right. The real question is whether the open-source community is prepared to fight the next war. The battle is not about code. The battle is about the law. And in the law, the ones who write the rules are the ones who win. Anthropic is writing the rules. The open-source community is hoping that the rules are not too restrictive. That hope is not a strategy. The AI industry is in its 2020 DeFi Summer. Everyone is building. But the people who are building the regulation are the ones who will be here in five years.\n\nThe regulatory capture accusation is a signal, not a solution. It tells us that the narrative of "safe AI" is a weapon. The question is, who is wielding it? The answer is clear. It is the ones with the deepest pockets. The lesson for the open-source community is to stop being naive. They need to build their own lobbying power, or they need to accept that their future is defined by the corporate giants. The choice is theirs. History does not repeat itself, but the structure of power is remarkably consistent. The audit is done. The risk remains.