Safety Regulation as a Competitive Moat
The line between protecting the public and protecting your market position is thinner than most executives admit. Anthropic, the AI startup that is now valued at nearly $1 trillion, has positioned itself as the industry’s most vocal advocate for state-level regulation. The company supported transparency laws in California and New York last year, then pushed for third-party auditing requirements in Illinois, and recently endorsed a Massachusetts policy that empowers the state attorney general to seek injunctive relief against non-compliant AI labs. Each new law raises the bar for what counts as responsible AI development. Each new requirement adds another layer of cost and complexity that only the largest players can absorb comfortably.
The Regulatory Gap That Creates Winners
The gap between what lawmakers understand about AI and what the technology can actually do is widening faster than any legislative session can close. Anthropic’s head of US state and local government relations, Cesar Fernandez, told WIRED that transparency and self-reporting are “no longer sufficient safety measures for the most powerful AI systems.” [2] This admission from a company that helped write those very transparency laws reveals something important: the rules that seemed rigorous in 2025 already look like training wheels. The company is now pushing for measures that would require independent safety audits and give state attorneys general the power to block deployments they consider unsafe.
But here is where the strategy gets interesting. Fernandez insists that Anthropic only supports bills applying to “large AI model developers” — companies that have spent hundreds of millions on development and generate more than $500 million in annual revenue. “It’s hard to imagine a startup meeting that threshold,” he told WIRED. This framing conveniently excludes every early-stage competitor while wrapping the company in a cloak of public responsibility. The definition of “large” is elastic enough to capture any startup that raises significant funding, yet narrow enough to leave Anthropic itself comfortably inside the regulated zone.
The Capture Accusation That Won’t Die
David Sacks, the former White House AI czar and current technology adviser to President Donald Trump, has claimed that Anthropic is essentially trying to get cumbersome laws passed to trap smaller AI startups in red tape. Critics argue that Anthropic is essentially trying to get cumbersome laws passed to trap smaller AI startups in red tape. The accusation carries weight because it describes a pattern of behavior that is well-documented in other industries. Companies that cannot beat their competitors through innovation alone often turn to regulation as a competitive moat.
Fernandez denies the charge, arguing that any company large enough to develop powerful AI should face the same rules because the underlying risks are identical. He told WIRED that part of Anthropic’s goal is to “inspire a race to the top developing the most safe and secure AI systems.” The logic sounds reasonable until you examine who gets to define “safe” and “secure” in the legislative language. Anthropic has been spending years warning lawmakers about catastrophic AI risks — financial disasters, mass deaths, systemic failures. These warnings create the emotional foundation for regulation. Then the company shows up with proposed language that happens to match its own operational capabilities.
The Federal Line That Protects State Power
Anthropic draws a careful distinction between what states should do and what should remain federal territory. In a policy document published last month, the company recommends that governments should have a mechanism to block companies from deploying unsafe AI models. But Fernandez notes that this power should belong to the federal government, not state lawmakers. This position seems reasonable until you consider the current reality: Congress has stalled on passing any meaningful AI regulation, and the Trump administration recently told Anthropic to suspend access to its two most powerful models for foreign nationals.
The company was not happy about that move. In a blog post following the export control directive, Anthropic argued that blocking AI model deployments should only happen through a fair, transparent evaluation process. The company wants the power to exist at the federal level, but only if the process meets its standards. This creates a situation where Anthropic can simultaneously advocate for state-level regulation that binds its competitors while reserving the right to complain when federal power inconveniences its own operations.
The Open Source Question Nobody Answers
The most revealing aspect of Anthropic’s regulatory push is what it does not say about open source AI. Fernandez told WIRED that the company has not specifically targeted open source models in any legislation it has endorsed. “It’s less of a question of the model construction and more of the model capabilities,” he explained. This distinction matters because open source AI models represent the most direct threat to Anthropic’s business model. If anyone can download, modify, and deploy a competitive AI system without paying licensing fees, the trillion-dollar valuation starts to look fragile.
Some AI researchers have dismissed these claims as regulatory capture by another means Some AI researchers have dismissed these claims as regulatory capture by another means. They argue that Anthropic’s real goal is to persuade the US government to ban open-weight Chinese models, which could prompt thousands of American businesses that rely on them to turn to Anthropic’s enterprise offerings instead.
The Job Displacement Question That Goes Unanswered
The concerns that actually worry American voters — losing jobs to AI, data centers arriving in their communities, chatbots affecting their children — have not inspired a comparable legislative campaign from frontier AI labs. Anthropic and several rivals have promised that ordinary taxpayers will not be stuck paying for data centers. The company has published proposals for responding to future AI-driven job displacement. But these commitments lack the specificity and enforcement mechanisms that Anthropic demands for safety regulation.
This asymmetry reveals the real priority. Safety regulation that targets catastrophic risks — financial disasters, mass deaths — sounds urgent and important. It creates headlines and generates political momentum. But it also happens to create barriers to entry that protect established players. Job displacement regulation would require companies to account for the human cost of automation, which would directly impact their growth trajectories. Data center regulation would force companies to negotiate with local communities about energy consumption, water usage, and property values. These are the regulations that would actually constrain Anthropic’s business, and they are notably absent from the company’s legislative agenda.
The Historical Pattern That Predicts the Future
The playbook Anthropic is running has been tested in other industries. When Uber faced regulatory battles across the United States, it hired Cesar Fernandez to help win policy fights in state capitals. Fernandez now works for Anthropic, bringing the same approach to a different technology sector. The sports betting industry, where Fernandez also worked, provides another template: companies that cannot win through market competition alone often turn to regulation to limit the field.
The pattern works like this: identify a genuine public concern — safety, fairness, consumer protection — then propose regulations that address that concern while simultaneously disadvantaging smaller competitors. Frame the regulations as necessary for public good. Hire former regulators and policy experts who understand how to navigate the legislative process. Build coalitions with safety groups and labor unions that lend moral authority. Then watch as the regulatory burden creates a moat that protects your market position.
The Sincerity Question That Changes Everything
Whether Anthropic’s efforts are sincere or strategic matters less than the outcome. The company is playing a powerful role in shaping the future of AI policy. Its latest model releases brought the cybersecurity capabilities of advanced AI into the national spotlight. Before that, the company had already spent years warning lawmakers across the country about catastrophic risks. The warnings are not wrong — advanced AI does pose real dangers that deserve thoughtful regulation.
But the structure of the regulation matters as much as the existence of it. Rules that require third-party auditing, transparency reports, and compliance departments create fixed costs that only large companies can absorb. Rules that define “large AI model developers” by revenue thresholds create a regulatory cliff that startups must climb while established players sit comfortably on the other side. Rules that focus on catastrophic risks rather than everyday impacts create a narrative that serves the interests of companies building the most powerful systems.
The Choice That Lawmakers Face
State legislators now face a choice they may not fully understand. They can accept Anthropic’s proposed regulations, which come packaged with expert testimony, draft language, and political support from safety groups. Or they can develop their own regulatory frameworks, which would require understanding the technology deeply enough to write rules that actually protect the public without accidentally creating monopolies.
The first option is easier. It produces headlines about protecting the public from dangerous AI. It satisfies voters who want action. It aligns with the messaging of a company that has built its brand around safety. The second option requires work. It demands that lawmakers understand the difference between model capabilities and model construction. It requires them to see the competitive dynamics behind the safety rhetoric. It asks them to write rules that protect the public without protecting incumbents.
History suggests most lawmakers will choose the easy option. They will pass the regulations that Anthropic supports, celebrate their commitment to AI safety, and move on to the next crisis. The startups that cannot afford compliance will either fail or get acquired. The market will consolidate around the players that helped write the rules. And the public will be left wondering why the AI industry looks like every other regulated industry — dominated by a few large companies that helped design the regulations that protect them.
The Irony of the Safety Argument
There is a deeper irony in Anthropic’s position that deserves attention. The company’s founding mission is “to ensure that the world safely makes the transition through transformative AI.” To fulfill this mission, the company believes it must build a massive business based on developing and selling access to advanced artificial intelligence. This creates a fundamental tension: the company that profits from advancing AI capabilities is also the company that wants to regulate those capabilities.
Every regulation Anthropic supports creates a competitive advantage that makes its business more valuable. Every safety requirement it advocates for raises the cost of entry for potential competitors. Every warning it issues about catastrophic risks reinforces the narrative that only large, well-funded companies can be trusted with advanced AI. The company is building the regulatory infrastructure that will protect its market position while presenting itself as a public-spirited advocate for safety.
This is not to say the regulations are wrong. Some of them may genuinely improve safety. Third-party auditing, transparency requirements, and deployment review processes could prevent real harms. But the source of the regulation matters. When the regulated write the regulations, the public should pay attention to whose interests are being served.
The Future That Is Being Written Now
The next few years will determine whether AI regulation serves the public or the incumbents. Anthropic is pushing states to adopt tougher rules while Congress remains paralyzed. The company has the resources, the expertise, and the political connections to shape the outcome. Its competitors, particularly smaller startups and open source projects, lack the same advantages.
The question is not whether AI needs regulation. It clearly does. The question is who gets to write the rules. If the largest players dominate the process, the regulations will reflect their interests. If lawmakers develop independent expertise and listen to a broader range of voices, the regulations might actually protect the public.
Fernandez told WIRED that Anthropic’s goal is to “inspire a race to the top developing the most safe and secure AI systems.” That sounds noble. But races to the top have winners and losers. The winners in this race are likely to be the companies that helped design the track. The losers will be everyone else — smaller competitors, open source developers, and the public that depends on these systems without understanding how the rules that govern them were written.
The gap between what AI researchers publish and what ends up in products is closing fast. The feedback loop between system deployment and system bias is accelerating. The question of responsibility when a system gets it wrong is becoming urgent. Anthropic is offering answers to all these questions. The public should examine those answers carefully before accepting them.
Sources
1. Anthropic
2. WIRED
3. White House