Locke, Meet Claude

Property is fundamental to individual freedom and, by extension, one of the key purposes of government. John Locke made that much clear. A few excerpts from the Second Treatise of Government testify to his views, which warrant renewed attention for two reasons. First, Locke heavily influenced the Founding Fathers. On the 250th anniversary of their efforts, it’s worth keeping Locke’s perspective in mind. In the lead-up to July 4, rereading Locke is a great way to revisit and celebrate what our forebears fought for and sought to protect. 

There’s a more immediate reason to pay attention to Locke’s thinking. Artificial intelligence (AI) has emerged as a powerful new tool for expressing ourselves, building new products, and offering novel services. This new modality of labor has ignited legal skirmishes over the extent to which the government may or should interfere with the development and use of AI. 

Consider the ongoing row between the Department of Defense and Anthropic. Defense Secretary Pete Hegseth a Supply Chain Risk following Anthropic’s insistence that the government not use its models for domestic mass surveillance and the operation of lethal autonomous weapons, even if those uses would be lawful. Judge Rita F. Lin of the U.S. District Court for the Northern District of California stayed that designation. In doing so, Judge Lin opened the door for the company to continue its work with the federal government. 

Judge Lin’s conclusion rested on several prongs. She found that the evidence indicated that Anthropic was punished for publicly critiquing the government’s negotiating posture. She also determined that “[n]othing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government.” More generally, she condemned the idea that the power of the state should be used for the punishment or suppression of disfavored expression. Anthropic was also deemed likely to succeed on its challenges under the Fifth Amendment and the Administrative Procedure Act. 

It’s unlikely to be the last time questions arise around how the government may dictate private creation and use of AI. (Indeed, there is a separate court proceeding between DoD and Anthropic on this same issue but under a different statute.) Today’s AI is the worst AI we’ll ever use. Today’s rate of AI adoption is the lowest it will ever be. Today’s intensity of AI use is also the weakest it will ever be. For the foreseeable future, every day will mark a new high point in the number of people using increasingly capable AI systems for an ever wider range of tasks. That fact will inevitably invite government attempts to limit or otherwise direct that use. 

You, your friends, and your loved ones may currently be thinking that AI is fairly useless. You may scoff at the idea that accessing these tools would ever justify an extensive legal fight. After all, didn’t you read recently that AI couldn’t ? (Note that this shortcoming is no longer true of advanced models, but that doesn’t stop AI skeptics from citing it as evidence of its purported flaws). If that’s your stance, my hunch is that there’s little I’ll be able to do to change your mind. I’ll instead invite you to think of your children and their children. For better and worse, they will lean on AI tools from a young age and likely to their benefit. Tools like Amira (), Flourish (emotional-resilience AI app), and Waymo (autonomous vehicle) will increasingly help students learn, adapt, mature, and, to be blunt, thrive. Allowing excessive government interference with such tools risks delaying or even denying their full use by current and future Americans.

Before the government comes for our AI, now’s the time to consider how the first principles of our constitutional order may apply to safeguard our use of this novel tool. This exercise warrants much more analysis than a humble essay. There’s much work to be done on the nexus between access to AI tools and self expression, on the right of association, and the right to access information. My aim is merely to start this prophylactic exercise — building a layered defense against any attempted undue intrusions by the state. My hope is that others will join the effort and contribute to the surrounding AI use, building a wall of fundamental legal protections. To start, I’ll focus on AI development and consider it as an extension of our labor and, consequently, as a form of property. In anticipation of mapping out that argument in greater detail in further pieces, I’ll begin the process by pointing to four key lessons on property from Locke, briefly identifying how they relate to battles over AI. 

Lesson 1: AI tools will increasingly be a product of our own labor, lending support to the idea of personalized AI as a form of property subject to heightened protection 

Whatsoever, then, he removes out of the state that Nature hath provided and left it in, he hath mixed his labour with it, and joined to it something that is his own, and thereby makes it his property. It being by him removed from the common state Nature placed it in, it hath by this labour something annexed to it that excludes the common right of other men. For this “labour” being the unquestionable property of the labourer, no man but he can have a right to what that is once joined to, at least where there is enough, and as good left in common for others. 

Under Locke’s labor-mixing theory, AI tools will increasingly take on the attributes of property. We’re in the early stages of personalizing AI tools to our values, personal dreams, and professional goals. Tools today allow users myriad measures to tailor AI to their individual whims. The means of personalization will only increase with time. AI agents of the near future may have access to your emails, texts, files, docs, etc. (all with your consent). The resulting functionalities will neatly and specifically align with each user. As this individualization deepens, the bar for government intervention into AI access and use should increase. 

Lesson 2: AI tools carry liberty interests that likewise justify property protections. 

Man … hath by nature a power … to preserve his property—that is, his life, liberty, and estate, against the injuries and attempts of other men. 

As recognized by Judge Lin, the liberty interest protected by the Fifth Amendment covers “the right to follow a chosen profession free from unreasonable government interference.” She asserted that the government’s hasty and legally baseless designation of Anthropic as a Supply Chain Risk stopped the AI lab from offering its tools to many customers. This prohibition did not follow from a sufficiently detailed and thorough process. As states and the federal government consider other limitations on how and when people and private entities offer AI services, this case and Locke’s conclusion ought to come to mind. AI tools can be interwoven with fundamental and essential acts of labor. Economists predict a world in which many Americans may be able to launch entire businesses with a team of AI agents. In such a setting, it will be imperative that governments adhere to clear process requirements before inhibiting such uses.

Lesson 3: The government has a role to play in ensuring broad access to the AI tech stack 

As much as any one can make use of any advantage of life before it spoils, so much he may by his labour fix a property in. Whatever is beyond this is more than his share, and belongs to others. Nothing was made by God for man to spoil or destroy. 

Locke took issue with individuals amassing more property than they could make productive use of. This concern may map onto several domains of the AI tech stack. For one, there’s a dearth of quality training data. As massive studios abuse copyright law as a shield to prevent more exhaustive AI training, this dearth of data will only become more dire. As access to quality data increases, startups will find the barriers to entry grow higher and higher. This will not only concentrate power in the hands of a few firms but diminish the likelihood of new innovation. 

Notably, the Founders, perhaps channeling Locke, regarded intellectual property as a means to an end: the generation and diffusion of knowledge. This functional vision of property conflicts with that of those who seek to generate recurring rents from their hoards of data. As courts continue to debate whether the fair use exception of the Copyright Act of 1976 covers training data, Locke may be a useful citation. 

It may also be the case that single firms controlling ever more quantities of scarce computational resources become a Lockean concern. There’s little question that such compute will likely be used. The more important inquiry is whether there are sufficient means for researchers, scientists, and others to conduct basic R&D that has traditionally been essential to human flourishing. Given this tension, there’s a need to investigate what constitutes “spoilage” in the context of access to and control over computational resources. 

Lesson 4: The purpose of the government is the protection of property, defined broadly. 

Men being, as has been said, by nature all free, equal, and independent, no one can be put out of this estate and subjected to the political power of another without his own consent, which is done by agreeing with other men, to join and unite into a community for their comfortable, safe, and peaceable living, one amongst another, in a secure enjoyment of their properties, and a greater security against any that are not of it. 

Locke’s social contract is not a grant of power to the state. It’s a leash on it. Citizens trade a portion of their natural freedom for the assurance that their lives, liberties, and estates will be protected from interference, whether by fellow citizens or by the government itself. That bargain doesn’t dissolve when the property at issue is novel or poorly understood. A government that brands an AI company an enemy of the state for expressing disagreement—as the Hegseth Defense Department attempted—inverts the Lockean order entirely. So does a state legislature that restricts which AI tools its citizens may use, or a federal agency that dictates which AI services a private firm may offer, without clear statutory authority and rigorous process. The purpose of government is not to manage innovation. It is to secure the conditions for individuals to innovate freely. As AI becomes more deeply interwoven with how Americans work, learn, and live, that foundational obligation grows more urgent.

Conclusion 

None of this is to say that AI regulation is always illegitimate. Locke himself recognized that property rights carry limits and that the government has a genuine role in preventing the monopolization of resources the community depends on. The concern is not regulation per se. It is a regulation that outruns its justification by arriving before the evidence, foreclosing the technology before its benefits are understood, and insulating the powerful from competition that would otherwise discipline them. That is the pattern worth resisting.

The Anthropic case is an early and unusually clear example of government overreach in the AI context. Most future cases will be murkier. They will arrive dressed in the language of safety, of consumer protection, of national security. Those justifications will sometimes be real. Distinguishing genuine regulatory purpose from pretext, incumbency protection, or hostility to disfavored speakers will be the central jurisprudential challenge of the next decade. Locke’s framework won’t resolve every case. But it establishes the baseline: the burden falls on the government, not the citizen.

Legal scholars have a particular obligation here. The courts hearing these cases are not steeped in AI’s technical architecture. They will need doctrinal frameworks that translate first principles into workable legal tests. The labor-mixing theory, the liberty interest analysis, and the anti-hoarding logic of Lesson 3 can be developed into coherent doctrinal arguments that judges can apply without becoming engineers. That translation work is overdue.

This essay is an invitation. The argument sketched here is deliberately incomplete. What it offers is a direction and a starting premise: that the same constitutional tradition that has protected property, labor, and expression for two and a half centuries is up to the task of protecting the tools Americans will increasingly depend on to do all three. The question is whether enough people will take that premise seriously before the regulatory window closes.