As the US midterm elections approach, voters are voicing concern about AI. According to an NBC News poll of registered voters, 57% believe the risks of AI outweigh the benefits. A rising political cohort is particularly concerned. A Pew Research poll showed that 61% of adults under 30 say more AI in society will make people worse at creative thinking. A recent Quinnipiac poll showed that 74% of Americans think the government is not doing enough to regulate AI.

Can you blame them? The CEOs of the largest AI companies chose a curious tactic: scaring their prospective users into submission. “Use it or get left behind” is the narrative, buttressed by gleeful proclamations that AI will destroy whole industries and cultural institutions.

AI’s detractors are on to something – thus far, the cost of AI disruption is not worth the benefits. A recent Goldman Sachs study showed AI impact on productivity amounts to a rounding error. It’s become better known for ushering in a new kind of bureaucracy: “workslop.” Harvard Business Review defines it as LLM-generated output that creates the illusion of productivity, but that later has to be corrected.

Indeed, slop is AI’s chief export. And there’s a simple way for lawmakers to mitigate its harms: institute a small slop tax.

“Slop” was Merriam-Webster’s word of the year for 2025 – defined as “digital content of low quality that is produced usually in quantity by means of artificial intelligence”. Once you see it, it’s everywhere you look. Fake bands streaming on Spotify; absurd AI-hallucinated cooking recipes; Amazon is flooded with whole books written via LLM prompt. Even the once reliable Google search is not safe. The search giant’s “AI overviews” overwhelm legitimate web results with factually incorrect AI answers, offering users tens of millions of wrong answers per hour, according to a study.

With the Maga right and Silicon Valley joined at the hip, Democrats have a critical opportunity to set the tone on AI legislation. It’s a chance to win back a generation that is already fed up with AI’s false promises. Yet the policies proposed so far miss the mark. Bernie Sanders has proposed a “pause” on AI, borrowing from doomers’ prognostications of a sentient artificial general intelligence – a distracting fantasy more at home in science fiction. The long-discussed universal basic income (UBI) is too broad and libertarian to a fault. AI probably won’t wipe out all jobs, and even if it did, throwing money at consumers won’t fix the underlying rot.

Instead, the left can win on AI by focusing on the most obvious and disheartening cultural side effect of AI’s meteoric rise: AI slop.

Faced with a flood of cheap impersonations, anyone engaged in cognitive and creative labor – from journalists and musicians to designers and educators – exists in a precarious situation. By unleashing billions of low-effort facsimiles of human creativity and cognition, it draws precious resources away from human creators.

The solution lies in the simplest of legislative measures: a minuscule tax levied on the largest AI companies to restore balance to what has heretofore been a one-way extraction. As it stands, AI slop is a malicious manipulation of human cognitive labor and the institutions that support it – something akin to a cognitive pollutant. A “slop tax” would ensure robust institutional support structures for human creativity forced to compete in a sea of meaningless content.

The structure is straightforward. If your company furnishes or hosts generative AI content, you are subject to an annual levy of ~1%. This is a negligible sum given the “abundance” that these companies’ revolutionary AI models should produce. This revenue goes into a publicly controlled fund that distributes grants back to varying kinds to cultural institutions, artists and researchers – the very same groups the models used as training data. The five largest public AI companies – Nvidia, Google, Apple, Microsoft and Meta – alone represent $18tn of value. A mere 1% tax represents a windfall appropriated to cultural workers, support for bedrock cultural institutions (such as local radio and newspapers, publishers, and educational programs) and grants for groundbreaking research and creative projects.

Sam Altman has pushed us to rethink the social contract. OpenAI’s recent memo on public policy has even invoked the New Deal to help us reimagine a public backstop for human cognitive production. This should include the profiteers of AI slop paying back into the system that they have strip-mined.

Why the isolation of “slop” when AI is already responsible for a litany of other negative societal costs? To start, slop is unlike most other technological disruptions. Instead of a new tool for humans to leverage, slop’s primary value proposition is that it serves as a quick and cheap replacement for human creativity.

Second, with all the hype behind generative AI, some forget it’s just an offshoot of machine learning. Frontier research into AI and machine learning should march on, but attempts to replace the institutions that nurture our cognitive sovereignty have a steep political and cultural cost. A slop tax would help to limit these threats.

Slop’s fatal flaw is that it is an attempt to use AI to simulate cognitive labor in situations where there are diminishing returns for going too fast. AI slop is predicated on a bet that our civilization will accept an inferior imposter in place of human creation. Democratic institutions that serve the public good – schools, newspapers and museums, to name a few – exist to nurture human flourishing. They are not problems to be solved by AI optimization. While various forms of AI-induced anxiety are rising, this is also an opportunity: a small tax on the worst parts of the industry could unleash a cultural renaissance.

• Mike Pepi is a technologist and author who has written widely about the intersection between culture and the internet. His book, Against Platforms: Surviving Digital Utopia, was published by Melville House in 2025