The Great AI Heist:

Full Text

Tech companies are building the future on the backs of human creativity — without asking for permission or offering a dime.

Imagine you spent years perfecting a craft — painting, writing, composing — and then shared that work with the world. Now imagine a corporation quietly collected every piece you ever made, fed it into a machine, and used it to build a product worth billions of dollars. They never asked for your permission. They never told you it was happening. And when their product began to replace you in the marketplace, they called it progress.

This is not a thought experiment. It is the operating model of the modern artificial intelligence industry. According to Brito (2024), this extraction of creative work is happening right now, at a scale most people have not yet grasped — and the public is only beginning to understand what that means.

To understand why this matters, it helps to ask a question most users never think to ask: What, exactly, are these systems learning from? As generative AI tools become embedded in daily life — writing emails, designing marketing materials, generating code — the answer has become harder to ignore. Every article, every digital painting, every photograph, every line of code ever posted online has been harvested and fed into these algorithms. Chayka (2023) documents how this happened without the knowledge or consent of the people who created the work in the first place.

The stakes could not be higher. Artificial intelligence is advancing faster than our legal and ethical frameworks can follow, and the window for meaningful intervention is narrowing. The decisions we make today — about consent, compensation, and intellectual property — will define the relationship between human creativity and machine intelligence for generations. If we get this wrong, we will not simply have failed a few artists. We will have dismantled the economic foundation that makes human creativity possible.

The Human Cost of Machine Learning

To see what this looks like in practice, consider the visual arts. When a user asks an AI image generator to produce a picture "in the style of" a specific living illustrator, the system draws on thousands of that artist's copyrighted works. Those works were absorbed without permission during training. The original artist receives no credit, no royalty, and no notification. They are not even aware their work was used. Meanwhile, the company that built the tool profits from every query.

Nor is this an isolated problem. Writers, photographers, musicians, and coders have all discovered that their life's work has been used as raw material for systems that now compete directly against them. A growing wave of lawsuits from authors, visual artists, and news organizations reflects the same underlying grievance. As Vincent (2024) reports, the AI industry has built its fortune on an act of mass, unconsented extraction — and creators are finally fighting back.

In response, the tech industry typically invokes the doctrine of "fair use." The argument goes like this: training an AI on publicly available data is no different from a student reading books in a library. But this analogy does not hold. A student reads to learn and then creates something original. A corporation, by contrast, processes billions of data points to build a commercial product that directly replicates and replaces the work it was trained on. As Heikkilä (2023) argues, that is not education. That is appropriation at industrial scale.

What Must Change

The solution is not to halt the development of AI. These tools offer genuine benefits — in medicine, education, accessibility, and scientific research. The question, then, is not whether AI should exist. It is whether it should be permitted to exist on terms that are fundamentally unfair to the people whose work made it possible.

At minimum, three things must change. First, AI companies should be required to obtain consent before using copyrighted work for commercial training purposes. Second, creators whose work is used should receive compensation — whether through licensing agreements, collective royalty pools, or other mechanisms yet to be designed. Third, AI-generated content should be clearly labeled, so that audiences can distinguish between human and machine-made work and make informed choices.

None of these demands are radical. They are the same protections we extend to every other form of intellectual property. The only reason they do not yet apply to AI training data is that the technology moved faster than the law. The companies that benefited from that gap have spent considerable resources ensuring it stays that way.

The creative class built the internet. Their words, images, and ideas are the reason these AI systems have anything to learn from at all. They deserve a seat at the table — not as an afterthought, but as the foundation on which this entire industry rests. The question is not whether we can afford to compensate them. It is whether we can afford not to.