America's AI Ambition Collides with Public Doubt and Ethical Gaps

Are we moving too fast? That's the resounding sentiment from the American public when it comes to artificial intelligence. A recent AP-NORC poll reveals a striking 64% of Americans believe AI is developing “too fast,” a sentiment echoed by a YouGov survey. This widespread apprehension stands in stark contrast to the determined march toward AI dominance touted by political leaders and the rapid, sometimes perilous, integration of AI into critical sectors, particularly the military.

While experts often foresee positive impacts on jobs—73% expect this, according to Stanford HAI—only 23% of the public shares that optimism, a significant 50-point divide. This chasm between expert confidence and public anxiety creates a volatile environment that demands immediate and thoughtful policy responses.

The Military's Unchecked Ambition

The Pentagon has declared an "AI-first" approach to warfare, evidenced by substantial financial commitments. Since 2016, the Defense Department has allocated at least $75 billion to AI-driven programs, a figure that a Brennan Center report notes could be significantly larger due to secret projects or those with unclear AI involvement. For 2026 alone, the Pentagon requested $13.4 billion for autonomous systems, and the FY2027 President's budget included $58.5 billion for AI-related investments.

Much of this investment has flowed to data analytics giants like Palantir and AI-powered drone manufacturers such as Anduril. Palantir leads the Maven Smart System, an AI platform reportedly used in Iran, Iraq, Syria, Ukraine, and Yemen, which integrates Anthropic's Claude to accelerate target analysis and simulate battlefield scenarios. Anduril specializes in autonomous systems, whose drones can navigate hostile environments and potentially strike targets with minimal human input, leveraging proprietary AI.

However, this rapid adoption carries significant risks. AI systems, including foundation models like Claude, are prone to inaccuracies and can generate misleading analysis while sounding persuasive. The Maven system, for instance, reportedly identified a tank correctly only 60% of the time in good weather, dropping to 30% in snowy conditions. Such errors, in a military context, can have deadly consequences. Media investigations, as highlighted by the Brennan Center, have shown instances where the Israel Defense Forces reportedly failed to sufficiently corroborate AI-recommended targets, partly due to pressure for quick approvals. This directly threatens to displace human expertise and judgment in life-and-death decisions, jeopardizing both troops and civilians.

Ethical Dilemmas and Corporate Pushback

The ethical concerns are not just academic; they are leading to direct clashes between the military and tech developers. Anthropic, for example, sought to prohibit the Pentagon from using its Claude AI model in fully autonomous weapons capable of identifying and firing on targets without human input. The company also aimed to prevent Claude's use for mass surveillance on Americans through analyzing location records, financial information, and other commercial datasets. The Pentagon refused these restrictions, subsequently blacklisting Anthropic from defense contracting by designating it a "supply chain risk." Last July, the Pentagon nonetheless struck deals with Anthropic and other major AI companies like OpenAI, xAI, and Google to develop military applications of their foundation models.

The push for an "AI-first" approach has profound implications beyond direct combat. The military's AI spending also includes as much as $9 billion on data centers and computing capabilities. Furthermore, AI enables mass surveillance, allowing the military to collect and piece together vast amounts of data—location information, social media posts—to recreate individuals' movements and associations at scale. When this generates sensitive insights about Americans, it raises serious privacy and civil liberties concerns, potentially undermining Fourth and First Amendment rights. The technology's inferences can be biased, flagging satire as security threats or associating protected characteristics with negative sentiments.

A Patchwork of Regulation Meets Public Demand

Despite the rapid military integration and the White House's promotion of AI to "win the AI race," the United States currently lacks comprehensive federal AI legislation. The federal government's approach has been described as cautious and pro-innovation, focusing more on supporting personnel and improving efficiency than on strict regulatory oversight. The Chief Digital and Artificial Intelligence Officer (CDAO) was established in December 2021 to integrate and scale data, analytics, and AI efforts across the Defense Department, but this primarily focuses on internal adoption and governance rather than broad public safety regulation. For example, the FY2027 President's budget request highlights that AI systems are designed to support personnel rather than replace them, with decision-making responsibility remaining with military leaders.

In contrast to the federal level, many U.S. states have introduced legislation addressing how government and public agencies use AI, creating a complex and fragmented regulatory landscape. This federalist complexity makes a unified national AI policy difficult to achieve. This regulatory vacuum stands in stark opposition to public sentiment. A very large bipartisan majority of 78% of Americans favor the government having the authority to audit AI programs that make decisions which significantly impact people's lives.

The growing chasm between public apprehension and the rapid, largely unregulated, deployment of AI in critical sectors like defense demands a more deliberate approach. Without robust regulation and clear ethical guidelines, the pursuit of AI supremacy risks undermining public trust and creating unforeseen, potentially catastrophic, consequences. We must bridge this gap with thoughtful policy that prioritizes safety and accountability over unbridled speed.