The New Gold Rush: Taxing AI's Extraction of Human Knowledge

The Industrial Revolution was powered by the extraction of coal and iron. The information age was built on silicon and rare earth minerals. Now, the AI era is fuelled by a different kind of resource: our collective human knowledge.

The central premise of most modern AI, particularly large language models (LLMs) like ChatGPT, is a colossal and uncompensated consumption of data. These models are not just “learning” from the internet; they are ingesting a vast, unstructured, and often unprotected resource—our books, our art, our code, and our conversations—to create something of immense commercial value. This is a form of digital resource extraction, and like its physical counterparts, it is high time it was taxed.

The “Data Mine” of Humanity

Historically, governments have regulated and taxed the extraction of natural resources to ensure that the wealth created from a shared, finite commodity benefits society as a whole. A tax on oil extraction, for instance, helps fund public services, infrastructure, and environmental remediation. Why should the same principle not apply to the digital gold rush?

AI companies have built multi-billion-pound empires by scraping the creative and intellectual output of millions of people without permission or compensation. This is not about stifling innovation; it's about rectifying an ethical imbalance and a massive, untaxed value transfer. The data is not a free-for-all; it is the product of human labour, creativity, and emotion. It is our collective digital heritage, and it is being used to automate away the very jobs and industries that created it.

A Tax That Reflects Real Value

This is not a simple “robot tax” on a piece of machinery. This is a sophisticated tax on the core business model of generative AI. Such a tax could take several forms, all designed to ensure that the creators are compensated and society benefits from the wealth their work helps generate:

A Model Training Levy: A one-off or recurring fee based on the size and scope of the training dataset. This would directly link the tax to the scale of the “extraction.” For a model trained on a billion web pages, the levy would be significantly higher than for a smaller, more specialised model.

A Data Usage Fee: A small, automated tax on every query or API call to an AI model. This would function like a royalty, providing a constant stream of revenue. Think of it as a micro-tax on the value being derived from the trained data with every interaction.

A Resource-Based Royalty: A percentage of an AI company’s revenue or profit, directly linked to their reliance on third-party, publicly available data. This is a way of saying, “If you are monetising the human resource, you owe a percentage back to the human collective.”

The revenue from such a tax could serve a vital, restorative purpose. It could be channelled into a fund to support the very creators whose work is being used—artists, writers, musicians, and journalists. It could also be used to address the societal challenges brought about by AI, such as worker displacement and the need for new education and reskilling programs. It would be a tangible way to ensure that the AI revolution serves not just a few tech giants, but all of humanity.

The Future of Taxation is Foundational

The conversation about taxing AI is no longer a theoretical exercise; it is a critical debate about our shared future. While we must address the carbon footprint of these new technologies, the deeper ethical issue is the extraction of our intellectual commons. The legal battles over copyright are just the beginning. The next frontier will be defining and taxing the most valuable resource of the 21st century: our collective knowledge. By doing so, we can ensure that the AI revolution is built on a foundation of fairness and equity, not on uncompensated exploitation.

The business model of most modern AI, particularly large language models (LLMs) like those developed by OpenAI and Google, is built on a foundation of “extracted” human creation. These models are trained on unimaginable quantities of text, images, and code scraped from the internet, which are, at their core, the product of human intellect, effort, and creativity. This is our collective legacy—the books we've written, the art we've created, the conversations we've had. And just like oil from the ground, AI is refining this raw material into something of immense commercial value, all without paying a cent for the source.