Please welcome back Lila Warren as a regular guest contributor to the 21st Century Tech Blog. Lila’s topic today takes on the copyright issue that artificial intelligence (AI) developers are facing in using material from the Internet to train their frontier generative AI models.
Training AI requires data. We are talking about massive amounts. The way AI companies do it is to screen-scrape millions of pages from the web, from digital renderings of books, articles, archives and more. Most of this content is protected by copyright. Using this content without permission infringes on the laws that govern copyright and protected materials. The result is that AI companies are facing a raft of legal challenges.
I’ll let Lila explain more.
The Complicated Impact AI is Having on Copyrighted Material
AI is transforming nearly every industry, and the creative world is no exception. From generating written content to producing original artwork, AI tools are reshaping how ideas come to life. But alongside these exciting possibilities comes a serious and growing concern: what happens to the rights of creators when machines are trained on their work? The relationship between AI and copyright law is complicated, contentious, and still very much unresolved.
How AI Systems Learn from Existing Content
At the heart of the copyright debate is how large AI models are actually built. These systems train on massive datasets that include books, articles, images, music, and other creative works pulled from across the Internet. In many cases, the content is used without the explicit permission of its original creators.
The AI doesn’t copy these works. What it does is learn patterns, styles, and structures from the material it accesses for training. This raises valid questions about whether using the material in this way constitutes unauthorized use. It’s a foundational problem that colours every other aspect of this debate.
The Gray Area of “Fair Use”
One of the most hotly debated legal concepts in this conversation is “fair use,” a doctrine under U.S. copyright law. AI technology companies argue that training their frontier models on publicly available data falls within “fair use” because the way the data gets used is transformative in nature.
It gets more complicated when you bring in laws governing copyright from other jurisdictions. For example, “fair dealing” is a comparable but more narrowly defined term applied to copyright and used by the UK, Canada and other Commonwealth countries. In the European Union, Singapore and Japan, they use “TDM exceptions,” which stands for “text and data mining.” Although not perfectly interchangeable, the goal of all of these legal definitions is to achieve the same for creator content protected by copyright conventions that date back 1710, when the first statutes were enacted in the United Kingdom.
Many creators and legal scholars strongly disagree with AI companies and their transformative claim that makes the issue of copyright less relevant. The legal arguments point out that the scale and commercial application of AI training goes far beyond what fair use was ever designed to protect. Courts are now beginning to weigh in, and their decisions over the next few years will likely shape the entire landscape of AI development and intellectual property law.
When AI Output Closely Mimics Original Works
What happens when what an AI actually produces bears a striking resemblance to specific works produced by authors, artisits and musicians? If an AI tool is prompted to create “in the style of” a human, it can produce something that feels unmistakably close to the original.
Style itself isn’t copyrightable, but the legal system is trying to define a line between stylistic influence and substantive reproduction. Until clearer standards emerge, the gray zone continues to cause friction between creators and AI technology companies.
The Challenge of Attribution and Ownership
When a human creates art, ownership is relatively straightforward. But when an AI generates content, the question of who actually owns that output becomes surprisingly complex. Is it:
- The user who wrote the prompt,
- The company that built the AI model,
- Or does no copyright protection apply to it because no human authored it?
In the U.S., the Copyright Office has ruled that purely AI-generated works without “meaningful” human authorship are not eligible for copyright protection. What is meant by “meaningful” remains open to interpretation and the ambiguity isn’t going away anytime soon.
What This Means for Creators and Businesses
For individual creators and businesses alike, navigating this environment requires both awareness and strategy. When professionals pursue efficient executive branding, they depend on original, authentic content that resonates with a target audience and reflects genuine expertise.
When AI-generated material blurs the lines of authorship and originality, it quietly undermines trust and authenticity, two qualities central to building a credible professional presence.
Businesses that rely on content creation should be proactive about understanding how they use AI tools and ensure their practices respect the rights of original creators.
Legislative and Industry Responses
Globally, governments and regulatory bodies are taking this issue seriously.
- The European Union has introduced provisions in its AI Act that require transparency around training data.
- Individual U.S. states are exploring their own legislative approaches.
- Some major platforms and AI companies are starting to develop licensing agreements with content creators as a more direct way to address these concerns.
These are encouraging steps forward, but meaningful and enforceable global standards are still a long way off. The pace of legislation simply hasn’t kept up with the pace of innovation.
Creators deserve to see their works protected.
The advancement of AI shouldn’t come at the cost of creators’ rights.
The courts, legislators, and industry leaders need to write a new rule book that recognizes AI isn’t going away and without guardrails, the technology’s creators need to respect boundaries.
