The Leanpub 60 Day 100% Happiness Guarantee
Within 60 days of purchase you can get a 100% refund on any Leanpub purchase, in two clicks.
See full terms...
AI Watermarks
How a Model Signs What It Writes
Every sentence an AI writes can carry a hidden signature, and the world's biggest models are already signing their output. AI Watermarking shows you exactly how it works, using a four-word vocabulary and arithmetic you can check with a pocket calculator. By the end you will know what a watermark can prove, what destroys it in seconds, and why short text can never be tested.
Minimum price
$10.00
$12.00
You pay
Author earns
PDF
About
About the Book
Every sentence an AI writes could be carrying a hidden signature. This book shows you how, with nothing more than a pocket calculator.
Text has no thickness. A paragraph is just characters, and nothing in it records who wrote it. Yet the world's largest AI systems are already signing what they write, invisibly, inside the words themselves. Google has been doing it in Gemini since 2024. The EU now requires it. Most people have never heard of it.
How a Model Signs What It Writes explains text watermarking from first principles, in plain English, with a vocabulary of only four words so that every number in every figure can be checked by hand. No code. No formulas you cannot follow. Just add, multiply, take a remainder, and watch a watermark appear.
You will follow Rosa, a lecturer who marks essays and cannot tell which ones a machine wrote, as she learns:
- Why guessing from writing style fails, and who it fails against
- How a language model actually picks each word, and where the hidden slack lives
- How a secret key, a look-back and a hash turn that slack into a signature
- Writing one watermarked word from start to finish, every step worked with real numbers
- How a checker keeps score without ever reading the text
- Why 40 words prove nothing, 100 are suggestive, and 300 are overwhelming
- What survives light editing, what a free paraphrasing tool destroys in seconds, and why short answers and code cannot be marked at all
- What SynthID-Text does differently, and what the EU AI Act's Article 50 asks for
With 43 figures and a running worked example, this is a short book you can finish in an evening and understand completely.
Who it is for
Teachers, editors and examiners who need to know what AI detection can and cannot prove. Journalists, policy makers and compliance teams facing new transparency rules. Developers and students who want the mechanism, not the marketing. And any curious reader who has wondered whether the text in front of them was written by a person.
What you will take away
A watermark is not proof of authorship. It is one line of evidence: very hard to forge, very easy to destroy, and useful only when you know exactly which question it answers. After this book, you will.
Feedback
Author
About the Author
Ritesh Modi
Ritesh Modi is Head of AI at MarketOnce and a former Forward Deployed Engineer at Microsoft. He has spent more than a decade building and shipping production systems across cloud, distributed computing, and applied machine learning, working with organizations ranging from global enterprises to fast-moving startups. His recent work focuses on applied large language models, designing systems that turn pretrained models into reliable, task-specific tools.
Ritesh has authored multiple technology books and speaks regularly at industry conferences on AI, cloud architecture, and software engineering. His writing philosophy rests on a simple belief: the best technical books are written by practitioners who still remember what it felt like to not understand something, not by experts who have forgotten. Every explanation in this book was tested against that standard, if it would not have made sense to him when he was first learning this material, it was rewritten until it did.
He writes, shares ideas, and connects with readers at www.riteshmodi.com. When he is not writing or building AI systems, he can be found mentoring engineers, exploring new architectures, or debugging a training run that should have converged three hours ago.
Earn $8 on a $10 Purchase, and $16 on a $20 Purchase
We pay 80% royalties on purchases of $7.99 or more, and 80% royalties minus a 50 cent flat fee on purchases between $0.99 and $7.98. You earn $8 on a $10 sale, and $16 on a $20 sale. So, if we sell 5000 non-refunded copies of your book for $20, you'll earn $80,000.
(Yes, some authors have already earned much more than that on Leanpub.)
In fact, authors have earned over $15 million writing, publishing and selling on Leanpub.
Learn more about writing on Leanpub
Free Updates. DRM Free.
If you buy a Leanpub book, you get free updates for as long as the author updates the book! Many authors use Leanpub to publish their books in-progress, while they are writing them. All readers get free updates, regardless of when they bought the book or how much they paid (including free).
Most Leanpub books are available in PDF (for computers) and EPUB (for phones, tablets and Kindle). The formats that a book includes are shown at the top right corner of this page.
Finally, Leanpub books don't have any DRM copy-protection nonsense, so you can easily read them on any supported device.
Learn more about Leanpub's ebook formats and where to read them
Write and Publish on Leanpub
You can use Leanpub to easily write, publish and sell in-progress and completed ebooks and online courses!
Leanpub is a powerful platform for serious authors, combining a simple, elegant writing and publishing workflow with a store focused on selling in-progress ebooks.
Leanpub is a magical typewriter for authors: just write in plain text, and to publish your ebook, just click a button. (Or, if you are producing your ebook your own way, you can even upload your own PDF and/or EPUB files and then publish with one click!) It really is that easy.