There is recent talk about taxing AI, which is probably a non-starter with policymakers adverse to doing anything to stand in the way of tech companies. But the tax idea isn’t totally crazy because AI these days is, for all practical purposes, a major tax on society.
Although AI has many beneficial uses, it remains unclear whether AI would be viable in the market if it weren’t artificially propped up by constant cash infusions by big gambler investors and by evading costs by sloughing them off on everyone else.
Unlike other technologies such as search, AI is tremendously expensive. There are many uses of AI that might not be sustainable in the market without venture capitalists pumping in massive investments. AI companies are shoving their AI in our faces at every turn, throwing it up against the wall in a frantic effort and hope that something sticks.
There certainly will be uses of AI that will provide more benefit than cost, but it’s hard to figure that out when there’s so much distortion of the market. Normally, the market would work as a mechanism of survival of the fittest, weeding out business models that don’t really work. But AI is growing today not because it is fit for survival; it’s being propped up.
AI imposes tremendous costs which operate as a tax on society. These are costs that AI companies are externalizing on all of us. Examples include:
Physical Resources
AI consumes resources such as water and electricity at massive levels. We all pay for this in higher utility prices, environmental impacts, and depleted water, among other things. It can affect farmers, and increase the cost of food.
Personal Data
AI is trained on our personal data, which companies scrape off the internet in violation of nearly every privacy principle and also in violation of many laws. I wrote about this problem with Woodrow Hartzog in The Great Scrape: The Clash Between AI Scraping and Privacy,113 Cal. L. Rev. 1521 (2025). At a minimum, it’s far from clear that AI companies are entitled to gather and use our data; just because it’s online doesn’t mean it’s free for the taking.
If companies couldn’t just hoover up our personal data and if they had to provide privacy protections for it, the cost of training their models would go way up and in many cases, they would have restrictions on their collection and use of data.
Intellectual Property
AI is also trained on our intellectual property. It’s far from clear that AI companies can just take whatever they want online and use it however they want. The tech company ethos is that if something is out there without iron walls around it, then they can just swoop in and take it. What if they had to pay for what they took?
Displacement and Disruption
AI also operates as a cost on everyone it displaces—all the creators whose livelihoods are undermined, all the workers it replaces. All these people are bearing the cost. Some tech leaders propose a universal wage to offset these costs, but it sounds like trickle down. I’m sure the universal wage, if it ever were to come to pass, would be a pittance, likely what people are paid for jury duty. And it won’t come to pass, because the one thing billionaires today hate the most is to share even a nickel with anyone else.
Labor to Clean Up Its Messes
AI also imposes problems that create significant labor for many people. In academia, schools must deal with the mess AI is causing with students who use AI improperly. Detection tools for AI use are extremely unreliable. A ridiculous number of hours are being spent dealing with the fallout. Journals for non-fiction and fiction are dealing with a tsunami of AI-generated submissions. They’re drowning. And in countless other areas of life, AI is creating headaches that are taking time, money, and resources to deal with. In many ways, AI is creating a nuisance that society must suffer from, like a toxic waste dump in all our backyards.
Harm
AI is causing real harm to many people, from deep fakes to encouraging self-harm to feeding delusional thinking. There are many tragic cases, a trail of tears and blood and death. As tech companies race along the superhighway to AI utopia, they ignore the gravestones along the side of the road. But these are real costs, real people, real lives.
* * *
All that said, I don’t hate AI. It has a lot of promise and has the potential to do a lot of good. Unfortunately, it has been unleashed on society in the worst possible way, like new neighbors who show up at your door to introduce themselves while their dog takes a big dump on your lawn.
It would be great if AI were to be rolled out responsibly, playing by the same rules as everyone else, with companies fully internalizing the costs. Instead, companies shirk the rules, lobby to undermine even the weakest of controls, fuel antidemocratic and socially destructive politics to further their aims, and slough off most costs on everyone else.
It’s high time AI were subjected to normal market forces, internalizing its costs. Then we’d see the areas where AI is truly promising.
* * * *
Daniel J. Solove is the Bernard Professor of Intellectual Property and Technology Law at the George Washington University Law School. He is the founder of TeachPrivacy, a company that provides workforce privacy, cybersecurity security, and AI training to companies and organizations around the world. He is the author of 10+ books and 100+ articles.
You can follow his events, writings, training, cartoons, and resources by subscribing to his free weekly newsletter.
Subscribe to Solove’s Free Substack
A supplement to Solove’s regular newsletter with more in-depth discussions
