Voters Wary of AI and Crypto as Industry PACs Flood Midterms

A new poll finds Americans deeply uneasy about artificial intelligence and cryptocurrency even as allied super PACs pour record sums into the 2026 midterm elections, raising the prospect of voter backlash against candidates tied to the industries. The survey, conducted for POLITICO, shows that skepticism could complicate the tech sector’s effort to translate financial dominance into lasting political power at a moment when experts warn that AI threatens to inflict even greater damage on democracy than social media already has.

The numbers are striking. Respondents strongly preferred candidates backed by groups calling for stricter AI regulation over those supported by PACs pushing for looser rules. A similar pattern held for environmental protection groups over crypto-aligned organizations. Only 12 percent of those surveyed believe special interests have too little influence over American politics, while 41 percent say they wield too much. The AI and crypto super PACs stand out even in an era of big money, sometimes rivaling or exceeding the fundraising totals of traditional party committees. Their spending spans both parties, complicating simple partisan narratives but amplifying fears that unaccountable tech money is shaping the rules that will govern emerging technologies.

Senator Chris Murphy, a Connecticut Democrat who has repeatedly called for robust AI oversight, sees an opening. “Democrats’ best approach is to make their spending an issue,” he told POLITICO. “People do not want AI companies to run them over culturally and economically. They don’t trust crypto.” Murphy’s assessment reflects a broader public fatigue with the way technology platforms have already reshaped American life, often in ways that feel corrosive to shared reality and institutional trust.

That corrosion is central to conversations taking place beyond the campaign trail. In a wide-ranging discussion with John Avlon, former Bloomberg executive editor Josh Tyrangiel argues that the social media era demonstrated how rapidly new technologies can destabilize democratic norms, and that AI represents the next, more powerful chapter. Tyrangiel, whose forthcoming book is titled “AI for Good,” urges a clearer-eyed view of the technology beyond Silicon Valley marketing. He points to concrete successes where AI has already strengthened government capacity rather than undermined it.

Operation Warp Speed, the Trump-era initiative that dramatically accelerated COVID-19 vaccine development, relied on AI-driven modeling and data analysis to compress years of work into months. More quietly, the Internal Revenue Service has deployed machine learning tools to detect fraudulent tax returns and improve customer service, producing measurable efficiency gains without fanfare. These examples, Tyrangiel contends, show AI’s potential to make government more effective at delivering public goods when deployed thoughtfully.

Yet such wins remain rare. Tyrangiel identifies bureaucratic inertia, risk-averse procurement rules, and cultural resistance inside agencies as formidable obstacles. Government, he argues, has largely failed to build the internal expertise and flexible processes needed to harness AI for public benefit. The result is a vacuum that private industry is only too happy to fill, often on terms that prioritize speed, profit, and minimal oversight over democratic accountability.

This tension sits at the heart of the current political moment. While Tyrangiel makes the case for smarter government adoption of AI, the midterm landscape reveals how industry is instead leveraging its resources to shape policy in ways that could entrench its advantages. The super PACs aligned with AI interests are not merely advocating for innovation; they are actively working to prevent the kind of stringent regulation that a plurality of voters appear to support. The poll suggests that if voters come to view these efforts as self-serving attempts to evade accountability, the financial advantage could become a political liability.

The pattern echoes the social media experience that Tyrangiel and others cite as a cautionary tale. What began as tools for connection morphed into engines of polarization, misinformation, and institutional decay. Platforms optimized for engagement frequently amplified the most extreme voices, eroded trust in expertise, and left democratic institutions struggling to respond. Many of the same companies and investors are now central to the AI boom, promising transformative benefits while resisting the very guardrails that might channel those benefits toward the public interest.

The 2026 midterms will test whether voters distinguish between the technology’s genuine promise and the concentrated power behind its deployment. Early signals from the polling suggest many Americans already draw that distinction. They appear open to AI’s practical applications, from faster medical breakthroughs to more efficient public services, but they distrust the motives of an industry that has repeatedly demonstrated how quickly it will prioritize growth over societal consequences.

For now, the advantage lies with those who can write the largest checks. But the poll indicates that translating money into votes may prove more difficult than anticipated. If candidates closely associated with the push for light-touch AI regulation underperform, it could mark an early rebuke to the idea that technological inevitability must dictate political outcomes.

The deeper challenge extends beyond any single election cycle. Rebuilding government’s capacity to steer powerful new technologies, as Tyrangiel advocates, would require overcoming the very bureaucratic and cultural barriers he describes. Without that work, the alternative is clear: AI will be shaped primarily by the same forces that bent social media to commercial logic, with consequences that extend far beyond any midterm tally. Americans’ unease, the poll suggests, is not a rejection of progress. It is a demand that technological change serve democratic ends rather than undermine them. Whether politicians of either party prove capable of meeting that demand remains the central unanswered question of the AI era.