The Coming AI Backlash
Artificial intelligence (AI) policy will be the most important issue in the 2028 U.S. presidential election. To most Americans, that statement probably seems absurd. AI policy was a total nonfactor in the 2024 presidential election. It does not even appear on Gallup’s latest tracking poll of the most important issues to American voters.
But AI capabilities are advancing exponentially and have only recently begun affecting daily life. The disruptive changes that AI will bring—to the economy, society, and national security—are coming far sooner than most people think. AI is already deeply unpopular. As those disruptions become personal to voters, they will generate political pressure that the current zero-regulatory environment cannot withstand.
Two studies independently concluded that the capabilities of the leading AI models have been doubling about every four months since 2024, and that rate is accelerating. Even if the current rate holds steady, by the 2028 election, the best AI models will be 250 times more capable than those of today. The implications of AI capabilities 250 times more powerful—let alone more than that—are difficult to comprehend. Their potential to accelerate scientific research, transform education, and create entirely new industries is enormous. But the disruptions will be equally profound and in many cases deeply personal for voters.
Rapid AI advancement will cause job losses. What is unclear is whether they will be confined to certain entry-level roles or sweep across all knowledge and physical labor, and whether displaced workers will find new jobs. OpenAI CEO Sam Altman stated that “there is going to be real pain” and “whole categories of jobs” will disappear. JPMorgan Chase CEO Jamie Dimon said that AI “will eliminate jobs,” adding that “people should stop sticking their head in the sand.” Even the optimistic scenarios in which AI causes only short-term job losses involve genuine hardship for people, and that hardship will generate a political backlash. Americans who lose their livelihoods to AI will not be mollified by macroeconomic arguments about productivity gains.
Powerful AI models also create potential for dangerous misuse. Modern AI tools are better than the best cybersecurity professionals at identifying critical vulnerabilities in secure software, potentially making them the world’s most powerful hacking tools. They can guide a person untrained in biology through engineering a dangerous virus. Those capabilities will only grow. Leading U.S. AI labs are now not publicly releasing their most advanced models due to fear of misuse, but that decision is purely voluntary; they are under no obligation to make it.
AI also gives the government the technical capability to monitor the activity of every American citizen using data purchased from commercial data brokers, such as cell phone locations, financial transactions, and online browsing history. That activity is not technically surveillance and requires no warrant. Previously, the sheer volume of data made comprehensive analysis infeasible. AI removes that constraint, at trivial cost. Americans expect the government to prevent the release of products that pose risks to public safety and will not support their government using AI to monitor their daily lives.
The national security implications of rapid improvements in AI are staggering. The U.S. military’s operations in Iran have put AI’s military applications on vivid display, enabling strikes at unprecedented speed and scale, and compressing target-identification from, as Commander of U.S. Central Command Admiral Brad Cooper, put it, “hours and sometimes even days into seconds.” AI will soon enable drones capable of fully autonomous combat operations. In future conflicts, the force deploying the better AI model will hold a decisive advantage. Maintaining the U.S. lead in AI is the priority on which all other national security priorities depend. Americans will not accept the United States military losing its technological edge, nor adversaries using U.S. AI technology to kill U.S. soldiers.
As this massive disruption looms, AI, although it is the driving force behind the U.S. economy, is deeply unpopular. Only 26 percent of voters have positive feelings about AI, compared with 46 percent who have negative views, and only 10 percent of Americans are more excited than concerned about the technology. That creates an unprecedented and strategically risky dynamic: the technology most important to U.S. economic, technological, and military competitiveness is also widely disliked domestically.
Calls for regulation are coming from both the left and the right. State lawmakers introduced over 1,200 AI-related bills in 2025 alone, including simple disclosure requirements, safety testing mandates, and outright moratoriums on new data centers. In March 2026, the White House released a national AI policy framework, but it focused on preempting state-level regulations while supporting targeted federal measures protecting child safety. Meanwhile, Senator Bernie Sanders (I-VT) and Representative Alexandria Ocasio-Cortez (D-NY) have called for a national moratorium on data center construction until strong regulations are in place. The current posture of near-zero federal regulation is politically unsustainable, as is a draconian pause in AI development. Substantial AI policy changes are coming; the only question is what they will look like.
Effective AI policy needs to address three distinct but deeply interrelated challenges simultaneously. First, domestic regulatory policies should ensure that advanced AI models are developed safely and deployed by the government in ways that do not undermine the fundamental rights of citizens. The government should be permitted to test them for safety before deployment, as it does pharmaceuticals and other sensitive technologies. And it should codify narrowly scoped limitations on the government using AI to surveil its citizens.
Second, social policies need to ensure that the economic returns from AI accrue to more than just a handful of technology companies and their investors. Without deliberate policy choices to share the gains broadly—through workforce transition programs, income support for displaced workers, and tax policies that capture a share of AI-driven productivity gains—the political backlash against AI will grow until it overwhelms any effort to maintain a permissive innovation environment.
Third, national security policies need to ensure the United States maintains global leadership in AI and maximizes its lead over China, which is the only viable long-term U.S. competitor. That requires not only a vibrant domestic AI innovation ecosystem but also robust technology protection measures that prevent China from using any U.S. technology—semiconductor manufacturing equipment, chips, cloud computing, or AI models—to advance its own AI ecosystem.
The consequences of failure on any single dimension—unsafe AI, extreme wealth concentration, or ceding leadership to China—would be catastrophic. The United States needs to get all three right. Crucially, those dimensions are not inherently in tension with one another. Effective technology protection measures that maximize the U.S. lead over China create the strategic space to regulate domestically and share the benefits broadly—without the risk of handing the advantage to China. Getting national security policy right makes it easier to get the other two elements right.
The instinct in Washington has been to resist regulation to protect U.S. competitiveness. But that instinct, however well intentioned, is self-defeating. The longer serious policymaking is deferred, the more political pressure builds. And the more pressure builds, the more likely the eventual response will be a hastily designed reaction to a crisis. The result could be the worst of both worlds: policies that are simultaneously insufficient to protect Americans and devastating to U.S. competitiveness.
The window for thoughtful, proactive policy is open now, but it will not last forever. AI capabilities are compounding, as is public frustration. AI will force itself onto the political agenda in 2028. When it does, the American people will expect policymakers to have serious answers ready.