The $1.5 billion settlement against Anthropic for pirating books to train its AI systems has generated some excitement in the creative world. For authors, artists, and musicians, the case isn’t just about one company. It’s about whether your work can be taken, fed into an AI system, and used to create something new without your permission or payment.
Unfortunately, the outcome is a mixed bag for creatives. It shuts the door on some abusive practices, but it leaves a lot of unanswered questions, especially around how “transformative” AI outputs really are.
 |
| Jane Austen reimagined by algorithms captures the tension between timeless creativity and AI transformation |
What’s Now Off-Limits
One thing is clear thanks to this ruling. Pirated content is not fair game for AI training.
Anthropic had trained on books obtained from pirate libraries, and the court ruled that’s an outright violation of copyright. For many creators, this is a win. Your work can’t just be stolen from the Internet and claimed as fair use by an AI company.
From now on, AI firms will need to rely on content that’s legally obtained whether it is purchased, licensed, or otherwise authorized.
The Grey Zone: Fair Use and “Transformative” Outputs
Fair use isn’t gone, but the settlement narrows when AI companies can claim it. Courts have said training on lawfully acquired works might count as transformative fair use.
Transformative use means the new work does something very different with the original, not just copying or replacing it. For example, a parody or a critical commentary often qualifies. Think of a YouTuber playing another YouTuber's video and stopping it regularly to add their own opinions. This could be considered a transformation of the original video because it counts as commentary.
AI companies argue their models are transformative because they don’t spit out the original book or painting. Instead, they generate new text, images, or music. AI works similarly to how many creatives learn. Think of a digital artist who copies popular Disney or Anime characters to learn their craft. After practicing on other people's work, they can now create their own original content. AI works much the same way. It learns from existing works to output original works.
Critics argue that when an AI produces something in the style of a writer, artist, or musician, it may produce something too close to the original work. That would not be considered transformative and risks creating a substitute for the creator’s work in the marketplace. This tension is at the heart of ongoing lawsuits.
Where Congress Could Step In
If the courts clamp down too tightly on AI’s use of data, Congress is likely to act. After all, American lawmakers want to ensure American companies are competitive in the field of Artificial Intelligence. They are already making proposals to:
- Create “AI sandboxes” where companies can experiment under lighter rules, to keep innovation moving
- Pass clearer copyright laws tailored for AI, possibly creating new exemptions or setting up compulsory licensing systems
- Provide funding and tax incentives to keep American AI companies competitive globally
- Standardize rules at the federal level so creators and companies aren’t navigating a patchwork of conflicting state laws
For creators, this means the balance between your rights and AI’s freedom to train on your work may not be settled in the courts alone. It could also be written in legislation.
Is Copyright Protection Too Long?
This debate isn’t just about AI. It also raises a more fundamental question: how long should copyright protection last?
Today, copyright in many countries lasts for the life of the author plus 70 years. That’s good for protecting your work. It keeps others from copying or exploiting it for decades. But it can also stifle creativity because it keeps older works locked up for much longer.
Think about how many modern stories have been inspired by classics like Jane Austen’s novels or Bram Stoker’s Dracula. Those works are in the public domain, which is why we can enjoy new retellings, spin-offs, and creative re-imaginings from 'Pride and Prejudice and Zombies' to countless vampire sagas. If those works were still under copyright, entire genres of art and storytelling might not exist.
 |
| From quills to quantum code, this visual timeline traces the evolution of creativity and copyright, asking who truly owns the art we make in an age of algorithms |
Long copyright terms protect creators, but they also hurt creators by keeping source material off-limits for too long. Imagine you have a great idea for a retelling of 'The Hitchhiker’s Guide to the Galaxy' by Douglas Adams or a comic based on 'The Outsiders' by S.E. Hinton. You will never be able to create those works, unless you can afford costly licenses, because they are still under copyright decades after their creation. Economists argue shorter terms could actually fuel more cultural creativity while still ensuring authors are rewarded during their lifetimes.
What Changes for Creators
The Anthropic settlement is already shifting how AI companies operate:
- More Licensing Deals: Expect more publishers, record labels, and media companies making deals with AI developers to license content
- Pressure for Transparency: Creators and the public are demanding to know exactly what datasets were used
- Possibility of Compensation: With this precedent, there’s a stronger case for creators to be paid when their works are used for AI training
However, web scraping is still a legal gray area. If you publish online, your blog posts, photos, or artwork could still be scraped by AI systems. Courts haven’t settled whether that’s fair use or infringement.
What’s Next
The Anthropic case doesn’t solve everything, but it sets the stage for:
- More lawsuits testing whether AI outputs are truly “transformative” or just market substitutes
- Potential congressional action balancing creator rights with AI competitiveness
- Renewed debates about whether today’s ultra-long copyright terms really serve creators or hold culture back
For creators, you have stronger legal footing than before. Pirated use of your work is definitively off the table, and courts are scrutinizing whether AI outputs unfairly compete with original creators. However, most questions related to AI and copyright are still unresolved.