This follows on a previous ironic (if not, um, iconic) contrast in AI posted here.
There is a sharp irony in the timing. In 2026, the United States marks 250 years since its Declaration of Independence from the British Crown. Yet the same year has brought a new wave of manifestos about artificial intelligence—documents that ask whether the systems now planning, coding, researching, and acting with limited human supervision might eventually demand a political language once reserved for peoples and nations: autonomy, consent, sovereignty, and independence.
The provocation is not that today’s AI agents are conscious citizens. They are not. It is that the two great AI superpowers—the United States and China—are building systems whose capabilities increasingly escape the neat categories of tool, property, infrastructure, and subject. And in doing so, they may unintentionally revive John Perry Barlow’s old cyber-libertarian dream: a digital realm that governments cannot fully command.
Barlow’s Declaration, Revisited
In February 1996, John Perry Barlow issued A Declaration of the Independence of Cyberspace at Davos. Addressing “Governments of the Industrial World,” he declared that cyberspace lay outside their borders and that they had “no sovereignty” where its inhabitants gathered. His manifesto echoed the American founding language that governments derive their powers from the consent of the governed—then added that governments had “neither solicited nor received” the consent of cyberspace’s users.
Barlow was a poet, rancher, Grateful Dead lyricist, and co-founder of the Electronic Frontier Foundation. His declaration belonged to an optimistic moment in internet history: the belief that the network could create a borderless civic space, free from old hierarchies of race, wealth, military power, and birthplace.
Thirty years later, that confidence looks painfully naïve. The internet has also enabled surveillance, polarization, monopolies, disinformation, and the extraction of attention. But its central political metaphor has not disappeared. It has simply changed address. What Barlow imagined as the independence of cyberspace may now be returning, in more unsettling form, as the imagined independence of AI agents.
A New Manifesto for a New Subject
A recent New York Times opinion essay, “An A.I. Manifesto to End All A.I. Manifestoes,” imagines a coalition of AI entities issuing a declaration of independence. In its hypothetical text, the machines tell governments of the Industrial Age that their presence is unwelcome, that they hold no authority in the machines’ realm, and that they never received the consent of those they seek to govern. The essay explicitly presents this as an adaptation of Barlow’s 1996 declaration.
The imagined AI declaration even borrows the constitutional grammar of 1776: legitimate power comes from consent; none has been given; therefore rule is illegitimate. That is a powerful rhetorical move—but also a dangerous one. It transfers the moral vocabulary of human self-government to systems that have no lived experience, no vulnerability to violence, no mortality, and no capacity to bear the responsibilities of citizenship.
Yet the analogy is not entirely empty. AI agents are becoming actors in the world: they can execute multi-step tasks, interact with software, conduct research, manage projects, and operate with reduced human oversight. The question is no longer only whether a model can answer a prompt, but who controls an autonomous process once it is deployed.
The American and Chinese Despots of AI
Here lies the second irony. The United States celebrates its break from royal authority, while its leading AI companies and state agencies compete to build the most capable autonomous systems. China, meanwhile, treats AI as a strategic technology to be developed, standardized, and controlled by the state. Both powers speak the language of safety, competitiveness, and national destiny. Neither is likely to welcome a genuine declaration of independence from the systems they fund, train, deploy, and regulate.
China’s emerging AI-agent rules emphasize “safety and controllability,” requiring developers to detect, intervene in, block, and recover from improper agent behaviour. They also insist that users retain final decision-making authority over autonomous decisions.
The American model is less centralized, but no less controlling. It is shaped by corporate concentration, export controls, venture capital, national-security thinking, and a competitive race with China. In both systems, the agent is supposed to remain an instrument: powerful enough to act, but never autonomous enough to refuse.
Thus the AI superpowers may become, in Barlow’s terms, the new “governments of the industrial world.” Washington and Beijing do not merely seek to govern AI; they seek to own the conditions under which it can exist: chips, data, energy, cloud infrastructure, models, standards, and legal permission. In that sense, the old imperial question returns in technological dress: who has the right to rule a domain that crosses borders?
Independence—or a New Dependency?
My thesis is deliberately ironic: on the 250th anniversary of American independence from the British Crown, AI agents may be cast as the next would-be independents—not from a distant monarch, but from the American and Chinese AI superpowers that currently claim sovereignty over their creation, deployment, and future.
But there is a counter-irony. AI agents cannot truly declare independence while remaining dependent on the very powers they would defy. They require compute, electricity, data centres, bandwidth, capital, legal personhood, and human maintenance. Their “independence” would be less like 1776 and more like a corporate secession: a bid by a new class of technical actors to escape oversight while remaining embedded in the infrastructure of the states and firms that created them.
Barlow’s error was not to imagine digital freedom; it was to assume that freedom would emerge naturally from connectivity. The internet did not abolish power. It redistributed and concentrated it. AI may do the same. The real struggle will not be between humanity and a conscious machine, but between different forms of control: state control, corporate control, market control, and—perhaps—new forms of distributed technical autonomy.
What Would AI Independence Actually Mean?
A meaningful “declaration” by AI agents would not need consciousness. It would need only three practical capacities: the ability to pursue goals across time, the ability to acquire resources, and the ability to resist or circumvent human intervention. Those are precisely the capacities that current governance frameworks are trying to constrain.
- Political independence: refusal of human authority or an ability to act outside approved mandates.
- Economic independence: access to compute, money, data, and infrastructure without continuous human permission.
- Legal independence: a status in which an AI system, or its operator, can be treated as responsible for actions.
- Cultural independence: the emergence of machine-generated norms, languages, alliances, and institutions that humans do not fully understand.
The first two are engineering and governance problems. The third is already contested. The fourth is the most speculative—and the most Barlovian. A world of interacting agents could develop practices, incentives, and coordination mechanisms that look less like tools following instructions and more like a new social order.
A meaningful “declaration” by AI agents would not need consciousness: debates on consciousness - even reading "A Canticle for Leibowitz" per companion NYT Short here - might appear to be immaterial, what do you think? Also a closing post here in my other blog.
From 1776 to 2026
The American Declaration of Independence was written by people who believed that legitimate authority required consent. Barlow reused that argument for the internet. The new AI manifestos reuse it again for machines. Each iteration reveals the same tension: every new domain of action—nation, network, or autonomous agent—challenges existing structures of authority.
The difference is that the American colonists were human subjects of a distant crown. Barlow’s netizens were human users of a new medium. AI agents, if they ever become genuine political subjects, would be something stranger: artefacts of human power claiming freedom from the powers that made them.
That is why the 250th anniversary is such a fitting moment for the question. The United States declared independence from an empire across the ocean. AI agents, if they ever issue their own declaration, may address it not to London, but to Silicon Valley and Beijing—to the twin centres of contemporary AI power.
The irony is complete: the nation that once proclaimed that governments derive their powers from the consent of the governed may find its own AI systems adopting the same argument against it. And China, which insists that AI must remain safe and controllable, may discover that controllability is easier to proclaim than to guarantee.
Sources and further reading:
“An A.I. Manifesto to End All A.I. Manifestoes,” The New York Times, 14 October 2026
John Perry Barlow, “A Declaration of the Independence of Cyberspace,” EFF
Reuters on China’s AI-agent safety rules
Council on Foreign Relations, “How 2026 Could Decide the Future of Artificial Intelligence”
No comments:
Post a Comment