AI and Copyright: What Every Author Should Know

Can AI-generated content be copyrighted? Can AI companies legally train their models on your published books without permission? If you use an AI tool to help write your novel, do you still own the copyright to it?

These are not hypothetical questions — they're live debates in courtrooms, copyright offices, and legislative chambers in multiple countries right now. And while the law is still catching up to the technology, authors need at least a working understanding of where things stand.

Important caveat upfront: this post is for general information only — it is not legal advice. The legal landscape in this area is changing rapidly, and specific situations require advice from a qualified intellectual property lawyer.

The Core Copyright Question: Can AI-Generated Work Be Protected?

In most countries, copyright law has historically required a human author. The requirement isn't explicit in every jurisdiction's legislation — it's often been established through court decisions and regulatory guidance — but the principle is consistent: copyright protects the creative expression of a human mind.

In the United States, the Copyright Office has been explicit: purely AI-generated content — content where there is no human creative selection or arrangement — does not qualify for copyright protection. They've reviewed this position in multiple registration decisions and it has been consistently upheld.

The picture is more nuanced for AI-assisted work — content where a human author makes creative decisions and uses AI as a tool. The Copyright Office has indicated that such work can be protected, but only for the human-authored elements. The extent of protection depends on the nature and degree of the human creative contribution.

In Australia, the Copyright Act 1968 requires a human author for most categories of copyright work. The Australian Law Reform Commission has reviewed AI and copyright, and while the law is not yet settled, the current position broadly aligns with the international trend: human authorship is required for copyright protection.

The Training Data Question: Were Your Books Used to Train AI?

This is the question most authors find most viscerally concerning — and with good reason.

Many large language models and image-generation AIs were trained on vast datasets that included published books, website content, and other creative works, often without the creators' knowledge or consent. Several major lawsuits are currently working through US courts, brought by authors and publishers against AI companies, arguing that using copyrighted works for training without permission infringes copyright.

The outcomes of these cases are genuinely uncertain. Some legal arguments hold that training on publicly available content constitutes fair use. Others argue that the reproduction of works during the training process — and the potential for AI outputs to reproduce substantial portions of training works — constitutes infringement. Courts are hearing evidence, arguments are being made, and decisions are pending.

What we can say with confidence: the answer is not yet settled, and the decisions made in these cases will significantly shape the relationship between AI companies and authors for years to come.

What Does This Mean For You, Practically?

Copyright of your own AI-assisted work

If you use AI tools to assist your writing — brainstorming, editing suggestions, marketing copy, research summaries — and you make the creative decisions, your work is almost certainly protected by copyright in the same way your unassisted writing is.

If AI generates substantial portions of the actual prose and you simply review and publish the output, the copyright position is considerably weaker. This is an area where understanding your platform's disclosure requirements (see below) intersects with your rights as an author.

Amazon KDP's disclosure requirements

Amazon now requires authors to disclose when content has been generated using AI tools. This applies to text, images, and translations. Non-disclosure of AI-generated content, if discovered, can result in content removal and account action. Whatever your view on AI, compliance with platform policies is not optional.

Other retailers and publishers are developing their own policies. Check the current requirements wherever you publish.

Protecting your own work

While the question of whether AI companies can legally train on your books is unresolved, there are practical steps worth considering.

Register your copyright where possible. In Australia, copyright is automatic upon creation — you don't need to register — but in the US, registration is required before you can sue for infringement, and timely registration allows you to claim statutory damages. If you publish in the US market, registration is worth considering.

Keep records of your creative process. Draft files, outline documents, brainstorming notes, revision history — these demonstrate human authorship and the progression of your creative work. This kind of documentation may become more important as copyright disputes around AI-generated content increase.

Some authors are also adding explicit "do not train" notices to their website and book content. Whether these notices are legally enforceable is an open question, but they establish clear intent, which may matter in future proceedings.

Stay informed

The situation is moving quickly. The Authors Guild, the Alliance of Independent Authors, the Society of Authors (UK), and the Australian Society of Authors are all engaged on these issues and publish regular updates. Following these organisations is one of the most practical things you can do to stay across developments that affect your rights.

Our Take

The law is catching up to the technology — and from what we can observe, it's slowly moving in a direction that takes author rights seriously. The lawsuits against AI companies for training data infringement are not fringe actions; they're being brought by major publishers and respected author organisations, and they're being taken seriously by courts.

In the meantime: document your creative process, comply with platform disclosure requirements, understand what protections exist in your jurisdiction, and get specific legal advice if you have a particular concern about your work.

Your words have value. Understanding how to protect them — even in a landscape that's still settling — is part of the business of being an author.

How are you thinking about copyright and AI in your own author business? This is a topic where the community benefits from open discussion — share your thoughts in the comments below.