IP Cash or Clash

LLM Hunger for Data, Copyright and the Global Pursuit of Equitable Value

We stand at a fascinating, yet precarious, intersection of technological ambition, creative rights and geopolitical strategy.

On one side, we have the insatiable hunger of Generative AI (GenAI) Large Language Models (LLMs) for vast quantities of data. On the other, the established principles of copyright protection. And swirling around both are global governments, vying to attract AI tech investment. This trifecta is on a collision course and the outcome will profoundly shape our digital future.

The core of the conflict is straightforward: GenAI LLMs, like ChatGPT, Gemini and their rapidly evolving peers, learn by processing enormous datasets. These datasets are often scraped from the open internet, a digital common rich with text, images, code and more. A significant portion of this data is, by its very nature, copyrighted. Authors, artists, journalists, musicians and developers all rely on copyright to protect their intellectual property and ensure fair compensation for their creations.

The prevailing argument from AI developers is that this "text and data mining" (TDM) constitutes "fair use" – that it's a transformative use of existing data to create new knowledge and capabilities, much like a human learning from a library. However, creators and rights holders vehemently disagree. They argue that their work is being exploited without permission or compensation, potentially undermining their livelihoods and the very incentive to create. Lawsuits from entities like The New York Times and GettyImages against AI developers are testament to this growing legal battle. The results so far are mixed: a US court found that training on lawfully acquired books can be fair use, yet Anthropic still agreed a $1.5 billion settlement with authors over pirated copies; Getty’s UK copyright claims against StabilityAI largely failed; and The New York Times’ case against OpenAI continues, leaving the definition of “fair use” in the age of AI unresolved.

This is where government policies enter the fray, adding another layer of complexity, often with a clear bias towards fostering AI development. Nations worldwide are locked in a fierce competition to become global leaders in AI, recognising its transformative potential for economic growth, innovation and national security. To attract investment from tech giants and foster a thriving AI ecosystem, many governments are exploring policies that favour AI development, often by offering exemptions or broad interpretations of copyright law related to TDM.

Across t he Atlantic, in the United Kingdom, similar tensions are playing out. The UK’sData (Use and Access) Act 2025 became law in June 2025 without the transparency requirements championed by the House of Lords, which would have forced AI companies to disclose what material they were using for training. In their place, the Act committed the government to a series of reports and impact assessments on copyright and AI, a process that is still unfolding. This move has drawn considerable dismay from the creative community. Even music icon Sir Elton John publicly expressed his anger, describing the government as "absolute losers" and asserting that if AI firms are allowed to use artists' content without payment, it would be "committing theft, thievery on a high scale." His powerful condemnation highlights the deep sense of betrayal felt by artists who see their livelihoods threatened by what they perceive as government-sanctioned exploitation.

The path of conflict is clear:

1. AI's Data Demand vs. Copyright's Core Purpose: Governments, eager to attract AI investment, are demonstrably willing to prioritise the needs of AI developers, often at the expense of creators' rights. This could lead to a "race to the bottom" where countries with the most permissive copyright laws become AI hubs, raising ethical questions about the balance between technological progress and fundamental human rights. The U.S. and UK examples vividly illustrate this governmental bias.
2. Regulatory Races vs. Ethical Obligations: LLMs need data, lots of it, to improve and innovate. Copyright law, in its current form, is designed to protect creators' rights to control and benefit from their work. These fundamental needs are often at odds.
3. Global  Disparity: Different nations are adopting varying approaches, creating a patchwork of legal landscapes. This lack of international harmonisation can create confusion, hinder cross-border collaboration, and potentially incentivise AI companies to operate in jurisdictions with the weakest copyright protections.

The resolution of this conflict will require careful navigation. It demands a balanced approach that fosters AI innovation while simultaneously upholding the rights of creators. This might involve new licensing models, collective bargaining agreements or even novel forms of digital attribution and compensation. Without a thoughtful and globally coordinated strategy, we risk stifling creativity, eroding public trust and building an AI future on a foundation of unresolved ethical and legal disputes.

The time for a clear, equitable path forward is now. Regulation is starting to demand one: since August 2025, the EU AI Act has required general-purpose AI providers to publish summaries of their training data and demonstrate copyright compliance, with enforcement powers applying from August 2026. Consent-based, rights-cleared data is becoming a commercial and legal necessity, not just an ethical preference.

This is the equitable path Mimo is building. Our priority registry is now open; register your interest on the Mimo website.

‍