Angela Hoy received the following inquiry from an author via email: “I am wondering if you have a special note on the copyright pages forbidding AI scraping.”
Angela wrote to me: “Lots of folks in the author groups are posting things like that, claiming that, if that’s in a book, it will stop AI programs from using their work. Some think that will give them a better chance at suing and winning later. I think that’s all total bunk. The AI companies don’t give a rat’s ass what’s on the copyright page. Whaddya think?”
UNSETTLED LAW
At first blush, I was dismissive of the efficacy of inserting such a “legal notice” on the copyright page of an author’s work. Obviously, existing federal copyright law has been no deterrent to any of the major tech players when they train their Large Language Models (LLMs) using hundreds of thousands of copyrighted works. In fact, many Tech CEOs knew that what they were doing was questionable (cutting, scanning, and disposing of copyrighted works in the name of training AI) and downright illegal (obtaining pirated ebooks to feed into their systems).
There was no plan for obtaining permission, nor for compensating authors. To date, even in light of the Anthropic settlement, AI oligarchs have been given a free pass. It’s disturbing at best. The Anthropic decision found that the scanning of print books, and uploading of copyrighted material was Fair Use under existing federal copyright law. The uploading of pirated copyrighted material was found to fall outside the bounds of the Fair Use doctrine, and a historical monetary settlement followed. Roughly half a million authors are scheduled to receive marginal compensation.
THE FEDERAL JUDICIARY IS SPLIT ON WHAT CONSTITUTES FAIR USE
On the federal level, the judicial branch of government begins at the district court level, proceeds to the appellate or circuit court level, and then to the Supreme Court, which promulgates the “law of the land.” The Supreme Court binds all the lower courts (circuit and district) in their decision making. Not to muddy the waters, but it is possible for states to afford broader protection under their own state constitutions or state laws.
The district courts are free to split on their decisions (i.e., opposing views on what constitutes Fair Use). In the United States, there are 94 federal judicial districts, including all 50 states, various territories and the District of Columbia. When these district courts disagree, it creates an appellate issue to be resolved by that state’s circuit court of appeals. There are 13 such courts in the U.S. When a split occurs among the circuits, the Supreme Court may step in and resolve the split among the circuits.
That is essentially what is happening with Fair Use rulings and litigation. Counsel and tech giants are waiting for splits to occur at the federal circuit level so that the Supreme Court may definitively define whether training Large Language Models with copyrighted works is, in fact, Fair Use under federal copyright law.
District courts do not bind each other within a Circuit. If you are in a state with multiple district courts, like California, one district court’s decision on Fair Use is merely persuasive precedent to another district court in the state. A decision by that state’s circuit court, however, is definitive. It binds the district courts within its circuit.
As I noted, a split among federal circuits (as they are not bound by each other) creates a path to the U.S. Supreme Court to settle an area of law such as Fair Use as it relates to AI training. Just as district court rulings are persuasive to another district court, circuit court opinions are persuasive, but not controlling, to another circuit court or district court outside that court’s particular circuit.
NO ONE IS IN UNIFORM AGREEMENT ON FAIR USE
The federal courts, the Department of Justice (DOJ), and the United States Copyright Office are not in agreement on what constitutes Fair Use. The DOJ couches its position in favor of Fair Use by expressing national security and international competition concerns (i.e., China). Currently, there is a four-prong test that the judiciary and Copyright Office harken to when determining what constitutes Fair Use. Therein lies the problem! We never contemplated Artificial Intelligence and its training when the four-prong test for whether something is fair use was first set forth in Section 107 of the Copyright Act of 1976.
The four prongs or factors in determining Fair Use are as follows:
- The Purpose and character of the use;
- Nature of the Copyrighted Work;
- Amount and Substantiality of the Portion used; and
- Effect on the Market
No one factor is controlling in determining Fair Use.
“The list is illustrative, not a checklist. Falling within a named category does not guarantee your use is fair. A professor who copies an entire textbook for a class still has to survive the four-factor analysis. The preamble opens the door. The four factors decide whether you walk through it.”
The test in application can be highly subjective. I take huge umbrage with the commercial application of copyrighted works to train and profit handsomely on the backs of authors. The court in the Anthropic case ruled that using copyrighted material – novels – had no impact upon the market for the author’s books. It also ruled that the use was more transformative and downplayed the commercialism aspect of training AI. That, in my opinion, is where the battlefront lines are drawn, and courts will differ greatly.
AI SCRAPING PROHIBITION
Adding an AI scraping prohibition to your copyright page may be worthwhile. Foremost, if your copyright protection is negated by the affirmative defense of Fair Use, an AI Scraping Prohibition can create separate legal causes of action in various jurisdictions. If your copyright claim defeats the Fair Use defense, such an AI scraping prohibition will bolster a willful copyright breach and can increase your statutory damages.
I would consider placing a scraping prohibition as a footnote or footer on every other Chapter’s first page. That footer may refer back to your Copyright page with more comprehensive AI scraping prohibitive language. This may ensure that the bot or company involved had actual or constructive notice. The nuances of this approach are covered in the Northwestern University Law Review.
Additional causes of action may include unjust enrichment, breach of contract (assuming you have set up a Terms of Use on your Copyright page), and misappropriation. A cause of action may also exist under the Digital Millenium Copyright Act (DMCA) § 1202 if the AI scraping involves the removal and destruction of copyright management information (CMI).
The New York Times took the AI Scraping Prohibition / Contract claim approach by modifying its Terms of Service to prohibit using its content “for the development of any software program, including … training machine learning or artificial intelligence (AI) system.” Bear in mind, the New York Times had a paywall, in addition to AI scraping prohibitions.
Bottom line: It’s a whirlwind right now. The New York Times case may well decide against the Fair Use defense to copyright infringement. In the alternative, the court may rule in favor of Open AI, but allow the case to proceed on the breach of contract / scraping claim.
It take a LOT of money to litigate these cases – which is one reason why the Anthropic case was certified as a class action (multiple plaintiffs similarly situated).
There’s no easy answer. It’s your call, but I recommend you need to ensure the likelihood that the bot or AI company doing the scraping has actual or constructive notice of your prohibition. My take is that you place a broad Terms of Use on the Copyright Page / Scraping Prohibition along with a footer interspersed throughout your novel. The copyright page alone isn’t enough.
It’s hard to fathom that it has come to this point for authors.
The AI companies don’t give a rat’s ass about your, your book, and your rights. They’ve made that abundantly clear. Anthropic’s settlement didn’t put a dent in its monetary coffers. Its scraping penalty was merely a pittance, and a proverbial cost of doing business.
RELATED
- Big Tech is Robbing Creatives and Others, and Sharing Data with Our Enemies! – by James M. Walsh, Esq.
- EPISODE 51: AI ABOMINATION! AI Companies are Buying Millions of Rare Books, Scanning Them, and Then SHREDDING THEM!!! – 07 2026
- SEND THIS LETTER to Your Representative(s) and Senators to STOP AI Companies from Destroying Old Books!
- Digital Dementia & Intellectual Atrophy: What AI is Doing to Children AND Adults – by Ric Forbes
- After Harming Children, and With THOUSANDS of Lawsuits Pending, Meta is Trying to Get an Exemption from Congress! – by James M. Walsh, Esq.
- BREAKING! In the Span of 2 Days, Meta Loses 2 BIG Lawsuits! And Now Thousands More Will Be Filed… – by James M. Walsh Esq.
- CRIMINAL CONSEQUENCES for Mark Zuckerberg? – by James M. Walsh, Esq.
- WHEN WILL IT STOP?! Meta (the owner of Facebook and Instagram) Now Accused of Profiting from Human Trafficking! – by James M. Walsh, Esq.
- Will Meta (Owner of Facebook & Instagram) FINALLY Get Prosecuted for Profiting from Scam Ads? – by James M. Walsh, Esq.
- Meta Removes Legitimate Legal Ads from Facebook and Instagram, But Allows Scam Ads to Keep Running! – by James M. Walsh, Esq.
- NEVER, EVER, EVER PAY USING A BANK TRANSFER…EVER!!! Here’s why… – by James M. Walsh, Esq. and Angela Hoy
- AUTHOR ALMOST LOST $50K!! We Saved an Author from a FAKE “Fox” Movie Deal, Which is Connected to a Known Publishing Scammer! By Angela Hoy and James M. Walsh, Esq.
- FRAUD ALERT: Why 5-Star Reviews on Trustpilot Can NOT Be Trusted By James M. Walsh, Esq.
JAMES M. WALSH, ESQ., is a former Navy JAGC officer and a recipient of the American Bar Association’s coveted LAMP Award for excellence in military legal assistance practice. A rolling stone, J.M. has globetrotted most of his adult life. After the military, J.M. pursued commercial real estate development, leasing, and asset management. He resides in Catania, Sicily. He spent almost twenty years in the Commonwealth of Pennsylvania’s Luzerne, Erie & Lackawanna Counties. His handiwork as an editor and author is interspersed throughout this novel. Leo A. Murray fondly refers to J.M. as his collaborative, literary ‘Coach’ or ‘Lieutenant.’
HAVE A QUESTION ABOUT SELF-PUBLISHING A BOOK?
a self-publishing services company that has been in business since 1998. Ask her anything.
ASK ANGELA!
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-israel.jpeg")
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-riley.jpeg")
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-pool.jpeg")
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-curtiss.jpeg")
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-clevenger.jpeg")
![](https://writersweekly.com/wp-content/uploads/2026/09/BLTestimonial-smith.jpeg")
[1] https://www.copyright.gov/fair-use/summaries/Bartz-v-Anthropic-PBC-787-F-Supp-3d-1007-ND-Cal-2025.pdf
[2] https://stanfordtechreview.com/articles/doj-ai-training-fair-use-openai-brief
[3] https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf
[4] https://fedlaws.org/section-107-of-the-copyright-act-1976-four-factors-of-fair-use/
[5] Id.
[6] https://www.culawreview.org/ddc-x-culr-1/nyt-v-openai-and-microsoft