Over a year ago, Donald Trump began his second term as president of the United States flanked by tech billionaires. Occupying seats typically reserved for family or past presidents were some of the world’s wealthiest men: Elon Musk, Jeff Bezos, and Mark Zuckerberg. Their physical proximity to the President would become more concerning in the following months, when the federal government’s lax — and oftentimes broadly encouraging — attitude towards relationships with tech billionaires seemed incorporated into favorable policy.
Musk, who would spearhead Trump’s government shed through the Department of Governmental Efficiency and become a close advisor with a contentious relationship to the President, had spent $200 million on the final stretch of the campaign. Bezos, CEO of Amazon, holds a contract with the federal government. Meta CEO Zuckerberg aligned his company with Trump’s priorities in the months prior to the inauguration. He also donated $1 million to Trump’s inauguration fund, along with Bezos and OpenAI CEO Sam Altman.
Mere weeks before, in his farewell address to the nation, former President Joe Biden warned about the “potential rise of a tech-industrial complex,” and the dangerous collusion of wealth and power left unchecked. The forming “oligarchy” of the few ultrawealthy, Biden said, threatened the very foundation of American democracy. In the scenes of the inauguration, it seemed clout and capital could push their way into Trump’s political agenda with minimal resistance.
Since January 2025, the prevalence of artificial intelligence (AI), in government, the private sector, and public life has rapidly expanded. Regulations — safeguards, guidelines, or protections of any kind — have not matched this expansion, at least at the federal level.
“I’m pretty skeptical [of successful AI regulation efforts], given the tenor and tone around this administration, [especially] when I saw multiple billionaires, most of them tech billionaires, standing behind the President when he took his Oath of Office,” says Graham Trainor, president of the Oregon Labor Federation, American Federation of Labor and Congress of Industrial Organizations (AFL-CIO). “As new AI powered technology is being unleashed on all of us, most of it unregulated and some of it dangerous, what we have been saying a lot is that, without common sense rules and a true commitment to boosting worker voice, there’s a future where good jobs, working conditions, safety, economic security, and worker and civil rights are at risk.”
Stefano Puntoni, a marketing professor and co-director of Wharton Human-AI Research at the University of Pennsylvania, comments that the mandates and constraints applicable to other digital products are applicable as regulations to AI. “I do think that AI is sufficiently different of a technology that it will require some specific regulation,” Puntoni says. “The shape of that is not very clear. “
Partially because of how quickly AI is evolving, and partially because of attempts at intentional regulation roadblocking, policy around AI has largely been lacking. Without existing protocols surrounding AI, enforceable accountability for AI — including the people behind it and those who champion its expedited implementation — doesn’t exist. Instead, “elected officials right now … are just trusting these tech billionaires to act responsibly, when we know that they’re failing to protect people from harm since guardrails are not in place,” laments Trainor. Trusting those who benefit directly from the success of their products to constrain themselves should not be the method employed to create healthy relationships between technological innovation, a government, and its citizens.
“There’s a lot of excitement about how AI might be used to improve cyber defense,” explains Matthew Ferren, an international affairs fellow at the Council on Foreign Relations. Potential integration of AI for cyber defense could look like using it to rapidly scan and identify vulnerabilities in software and help fortify software by rewriting software programs in difficult-to-compromise programming languages. Ferren adds, “All of this is happening very rapidly and is very new. There’s a lot of hype, … but I think most people are still watching and waiting to see exactly how that’s going to play out.”
As per the White House’s March 2026 National Policy Framework for AI, AI development should be done with the protection of children and investment in communities in mind, among other priorities, but without much other restriction. The policy says states should be as limited as possible on when they can regulate AI, and that regulation should be done by the federal government, which would invest in AI “by removing barriers to innovation, [and] accelerating deployment of AI applications across sectors.”
The current federal administration has maintained its rhetoric regarding free-range AI development unencumbered by state regulation attempts. A proposed version of the One Big Beautiful Bill Act included a 10-year moratorium on states enforcing or enacting AI-related laws — a provision supported by OpenAI and Google. The provision was struck from the version that passed in July. Then in December of 2025, an executive order instructed various members of the cabinet to challenge existing state AI laws.
A White House that wants to have the singular say on AI is dangerous. One that openly cozies up to Silicon Valley — and ventures to align their anti-regulation standpoints — even more so. “What tech billionaires really want is an unregulated landscape to deploy at their mercy and will, regardless of the harm and the costs, in the interest of innovation,” states Trainor. On March 25, 2026, Trump named new members to the President’s Council on Advisors on Science and Technology — a body to help shape federal policy — including big names at Meta, Google, and Oracle.
It looks like some of the loudest voices in the room all frame regulatory conversations as stifling innovation, and want to jump to the implementation of AI everywhere possible. Ferren says this impulse appears part of a larger trend over the past 20 years. “To some extent, … it has been very common in the tech industry … to just get the tech out there before anyone has a chance to figure out the downsides or how to regulate them,” he says. This has dire implications for the future; by the time the consequences are realized, “It will be so deeply embedded in economic and social structures that it’s too late to really do anything about it. Too bad for you, we’ve cashed our checks.”
Exactly what AI can offer governments is nebulous; it’s a general-purpose technology whose purpose depends wholly on how people choose to implement it. Our perceptions of realistic, pragmatic uses of AI are skewed by exaggerations of its capabilities. Especially in the context of government, AI carries mass risk — when involved with the systems that govern life for billions, why trust the selling points of untested tech? “Right now, everybody’s awestruck by the tech bros and the billionaires in Silicon Valley and what they are promising when, in fact, that’s very much divorced from reality,” says Trainor.
As Trainor sees it, reckless, unintentional implementation of AI is dangerous for working-class Americans, in particular. He draws a parallel to past pushes for task automation, which prioritized gains for companies and investors while negatively impacting working people. “There is an opportunity to avoid those past mistakes,” he says. “Unfortunately, this vision for a worker-centered technological future doesn’t reflect what we’re seeing today.”
Beyond challenging the lives of people in America, unregulated AI can bolster attacks against elements of democracy itself. From a cybersecurity standpoint, Ferren contends that AI tools might allow bad actors to scale up their campaigns with ease. Dr. Rachel George, a lecturer in International Relations at Stanford University, explains that AI risks undermining democracy in multiple ways, including “disinformation campaigns from deepfakes, manipulation of the information environment, and broader security challenges.” She sees AI as possessing potential for citizens to use positively, but highlights election integrity from AI-generated content as a risk.
As argued in an essay published by the Harvard Center for Ethics, authentication, transparency, and privacy are democratic principles undermined by AI developments. When AI regulation is prevented, and implementation is pushed hard, the American people lose influence on their government.
At the bedrock of American democracy is the idea that the people elect their leaders, and that by-nomination positions are filled in good faith and thoroughly vetted, typically through the Senate. The American people elect their peers as representatives, someone who can understand and elevate their voice, not programmed AI systems. Without conscious and deliberate allowances for AI presence, the possibility of AI being baked into and automating the democratic system is startling.
The human mind may be fallible, but it’s the devil we know — uses of technology to minimize human bias have already existed, but including new tech into those functions must be done carefully. The possibility of concentrating the power to use AI and decide how it is used in the government of an entire nation into the hands of the elite of Trump’s inner circle, shifts the power away from the people and threatens the democratic system.
When the military utilizes AI algorithms, it “is part of this long chain involved in identifying targets and potentially selecting them for a strike,” explains Ferren. Bias in any part of this chain has consequences, even if the algorithm itself is not explosive. “The example that’s getting a lot of press right now is the role that Anthropic’s algorithm may or may not have played in the military’s strike targeting systems and processes, and specifically the extent to which it was or was not responsible for an alleged strike on a school in Iran,” says Ferren. We must be able to judge the readiness and appropriateness of these systems before they are broadly incorporated into actions both consequential and routine.
Additionally, seemingly menial, repetitive tasks shouldn’t be automatically taken over by AI. If a congressperson responds to constituent letters using AI, Ferren exemplifies, “I think some of us would also say that, well, that sort of feels like cheating, and might break some of those human bonds that are kind of necessary to a healthy polity.”
According to a 2025 Gallup poll, 80% of Americans believe the government should prioritize maintaining rules for AI over rapid capability development. It’s vital “to voice and incorporate into policy this demand and this expectation that the vast majority of Americans believe should be the policy and the law of the land,” states Trainor.
State governments, which have thus far allowed for more diverse takes on AI regulation, must take proactive action to make tech policy based on accountability. Despite blocking attempts by the federal government, AI-related legislation has been proposed and passed across the country, all fights that should continue.
It isn’t that we shouldn’t implement newly developing tech, or that our government should be recklessly cynical, but rather that the American people — and every person impacted by the actions of the American government — deserve more than their leaders’ automatic trust in unproven systems. As the consequences of AI use mount, we need guardrails, and we need accountability.
