In the span of just a few years, artificial intelligence has gone from a futuristic curiosity buried in research papers to a daily presence in our offices, schools, hospitals, and even our dinner conversations. It can write essays, diagnose diseases, create art, predict our shopping habits, and drive our cars. Yet alongside this extraordinary promise runs an equally extraordinary anxiety. We worry about jobs disappearing, about false information flooding public life, about algorithms that make unfair and unknowable decisions about our mortgages, our insurance, and our freedom. Faced with these enormous stakes, one might expect governments to act quickly and decisively, the way they moved to regulate aviation, pharmaceuticals, or finance after early disasters. Instead, as Cecilia Kang, a technology reporter for The New York Times, has explained in her reporting, the effort to regulate artificial intelligence in Washington and in many democracies has slowed to a crawl. There have been hearings, white papers, executive orders, and grim warnings from experts, but no comprehensive federal law has been passed. The question of why is not simply a story about wonks and politicians; it is a deeply human story about fear, confusion, money, and the sheer difficulty of making rules for something that keeps changing shape.
To understand this paralysis, we have to begin with the strange nature of the technology itself. Artificial intelligence is not a single, stable thing like a drug or a power plant. It is an umbrella term for a sprawling set of methods, models, and products that evolve almost weekly. Large language models can be trained on enormous swaths of text, then fine-tuned, then connected to browsers and databases, then given voices and faces. A law written to address one tool is often obsolete by the time it is enacted. Regulators are not, for the most part, technologists. Many members of Congress, as Kang has noted, struggle to understand the very basics of how the Internet works, let alone the nuanced mechanics of neural networks, reinforcement learning, or algorithmic alignment. They depend on outside experts, but the experts themselves disagree. Some claim that AI poses an existential risk on par with nuclear war, while others insist such fears are overblown and distract from more immediate harms like bias, surveillance, and misinformation. When the people who build the technology cannot agree on what it is, what it can do, and what the most urgent risks are, it is almost impossible for a skeptical legislator to gather the confidence needed to impose rules. This is not simply ignorance; it is a structural mismatch between the pace of invention and the pace of governance. By the time a senator has learned enough about one model to ask a pointed question, the company has already moved on to another model ten times more powerful. The ground never stops moving, and so lawmakers, like children chasing a soap bubble in the wind, find themselves perpetually a step behind.
But there is another, less forgiving reason that regulation has stalled, and it has to do with power and money. The AI industry is not just a collection of clever start-ups in garages; it is dominated by some of the wealthiest corporations in human history. These companies have invested billions of dollars, and they see regulation as a direct threat to their market dominance. They have learned, as previous generations of tech companies did, that the language of safety and ethics can be a useful tool. In her reporting, Kang has shown how tech giants have positioned themselves not as opponents of regulation but as its most reasonable advocates, calling for “smart” rules and “responsible” innovation, while simultaneously deploying teams of lobbyists to limit what those rules would say. The gaps between public concern and private incentive are where legislation goes to die. A proposed law can sound commonsense in the abstract—say, requiring companies to disclose when content is generated by AI—but the fine print becomes a war zone. Lobbyists push for narrow definitions, broad exemptions, and softer penalties. They argue that overly strict compliance costs will hurt small businesses and American competitiveness. They fund think tanks and academic researchers who produce studies favorable to their cause. They open offices near the Capitol, hire former congressional staffers, and build relationships with the very people tasked with overseeing them. Meanwhile, ordinary citizens, whose jobs, privacy, and families are most at risk, have almost no direct seat at the table. The result is a profound imbalance. The people who profit from artificial intelligence are highly organized and well-resourced; the people who might be hurt by it are scattered, distracted, and often not even aware of the dangers until they materialize. Regulation is not just a technical problem; it is a political power struggle, and in that struggle, the machine owners have been winning.
Compounding this, the political environment in which AI regulation is being debated is perhaps the most polarized and dysfunctional in a generation. It is hard enough to pass laws about infrastructure or health care, where the facts are old and the stakes are clear. Regulating AI requires lawmakers to think about the future, to weigh hypothetical harms, and to share power with scientists and bureaucrats. Those are not conditions that favor action in a system defined by short election cycles and partisan posturing. In the United States, the issue has become entangled with broader culture wars. Some see AI as a tool of progressive censorship, others as a corporate tool of exploitation. Some want strict rules to protect jobs and truth, while others fear that any regulation will hand American advantage to China. The result is a chaotic patchwork. Some state governments have passed their own privacy and AI laws, while the federal government has relied on voluntary commitments from the industry—an arrangement that critics liken to asking the fox to guard the henhouse. Every time a horror story emerges, such as a chatbot encouraging a vulnerable teenager to harm herself, or a biased algorithm denying health care to Black patients, public outrage flares and congresspeople promise action. But after the news cycle fades, so does the urgency. The legislative calendar is crowded, committee jurisdictions are muddled, and the complexity of the issue makes it easy to defer. By comparison, the European Union has made notable progress with its AI Act, but even that landmark law took years, was watered down through intense lobbying, and still leaves many key details to be filled in by technical standards that do not yet exist. The American system, with its many veto points and partisan deadlock, is almost perfectly designed to stall. It is not that no one knows what to do; it is that the people who know are not the same people who decide.
Perhaps the most painful irony is that the public has never been more ready for change. Polls consistently show that large majorities of Americans are worried about AI, want more transparency, and support government intervention to protect privacy and slow job displacement. In a democracy, such sentiment should be a powerful force. Yet in practice, public concern has not translated into public action. Why? Because anxiety without a concrete directional target is hard to convert into policy. People know they are worried, but they are rarely asked what, exactly, they want regulated. They are given scary headlines and magical promises in the same breath, and the constant hype makes them cynical and numb. They hear that AI will cure cancer and also destroy democracy, and the contradictions do not inspire confidence in either technologists or regulators. This climate of confusion is fertile ground for the lobbying described above. It is easier to weaponize fear of regulation than to build support for it. Furthermore, the tech industry plays a crucial role in shaping the public narrative. It controls the platforms where citizens discuss the issue, the algorithms that determine what they see, and the news coverage that reaches them. When the industry says it is self-regulating, it can point to safety teams and ethical charters, even if those teams are being laid off behind the scenes. When critics demand government oversight, the industry warns of “innovation being slowed” and evokes the specter of missing out on the next economic boom. In such an environment, the issue ceases to be about protecting a baby in a bathtub and becomes a fight over identity, autonomy, and trust in institutions. Human beings do not like being told what to do by people they do not trust, and in America, we trust almost none of our institutions. The connection between public fear and legislative response has broken down. We are not too calm to act; we are too overwhelmed to be heard.
Yet despite all of this, there is still a path forward, and it begins by remembering that AI is not an unstoppable natural disaster. It is a set of choices made by human beings, and every choice can be governed. Cecilia Kang’s reporting suggests that the real breakthrough will not come from a single sweeping law, but from a slow accumulation of pressure: tireless journalism that exposes harm, local action that builds momentum, labor unions that demand protections, activists who push for transparency, and a public that refuses to treat complexity as an excuse for surrender. The goal does not have to be a perfect, utopian regulatory regime. It can be a set of practical and incremental measures: mandatory documentation of training data; independent audits of dangerous systems; liability rules that make companies responsible when their products cause harm; a national privacy law that lets ordinary people control what happens to their information; and a dedicated agency with enough technical expertise to keep up with the changing landscape. These are not radical ideas. They are the same kinds of safeguards we apply to every other powerful technology. But they will not happen automatically. They will require citizens to disengage from the culture wars long enough to demand something concrete from every candidate in every election. They will require politicians to spend less time asking robots questions at hearings and more time drafting bills that ordinary people can understand. And they will require a collective act of courage: to admit that the future is uncertain, that we do not know exactly which harms will materialize, but that we have an obligation to begin. The debate over how to regulate artificial intelligence is not just a policy dispute; it is a mirror of who we are as a society. It reveals how easily we are dazzled by power, how quickly we are paralyzed by fear, and how deeply we value the next breakthrough over the last safeguard. But it also reveals a capacity for care, for responsibility, and for the stubborn refusal to hand over the running of our lives to machines we do not understand. We have seen such stubbornness before—in the fight for food safety, for child labor laws, for environmental protection, for civil rights in the digital age. It is never easy, and it is never linear. Progress comes in fits and starts, through heartbreaking stories and incremental victories. The story of AI is still being written. Whether it becomes a tragedy or a triumph depends not on the technology, but on us.
