Scandals over artificial intelligence in professional writing are mounting, prompting a British startup to create a certification for books written entirely by humans. Books by People has secured approval from the UK Intellectual Property Office to issue a mark that indicates a published book has been authored by a person. The logo, designed to look like a thumbprint, will appear on book covers.
Esme Dennys, a cofounder of Books by People, told WIRED that readers should have the option to choose books written by people. She noted that while people are free to write or seek out AI books, those wanting human authorship should be able to find it easily. The company aims to create that choice.
The verification process
Earlier schemes, such as Human Authored, relied on an honesty system where authors promised they had not used AI. Books by People adds weight to its certification by requesting an authorship declaration. The company runs manuscripts through proprietary software and manually examines early drafts and research materials for signs of human input. Dennys stated the organisation looks at indicators that suggest a work was made during a natural process of drafting and redrafting.
Publishers will pay a flat membership fee plus a sum for every book submitted for certification. Five publishers have signed up so far, with the first book expected to carry the mark by the end of the year.
Industry context
This year has seen several scandals involving suspected AI use by authors. A literary agent cancelled a book deal reportedly worth $2 million. A publisher withdrew a novel from shelves after readers alleged a prize-winning short story was generated with AI.
Writers and publishers are not necessarily worried about being replaced by AI. Insiders say few people will ultimately want to read machine-generated books. Ted Chiang, whose 1998 novella Story of Your Life was adapted into the Academy Award-winning film Arrival, said knowing how something was created affects the experience. He described standing at the end of a conveyor belt watching stuff fall off the edge, which prevents emotional investment in whatever falls into your hands.
Repeated scandals could shake public confidence in the unspoken rule that every book on a shelf was authored by a real person who did not indulge in plagiarism. Sam Jordison, codirector at independent publisher Galley Beggar Press, which intends to adopt the Books by People mark, said publishing relies on trust between the publisher, writer, and reading public. He noted there is a feeling of revulsion about the idea of reading an AI text that purports to be human.
Support and criticism
The idea of a certification for human-authored books has received backing from various writers and editors. Markus Dohle, former CEO of Penguin Random House, described human storytelling as the lifeblood of the industry in a statement praising Books by People.
However, the routinization of AI detection is a double-edged sword for authors. They risk a false positive result spoiling their reputation. Tom West, CEO of the Publishers Licensing Services, warned that the organisation needs to be careful this does not become a witch hunt. He said the devil is in the detail.
New York-based startup Pangram, which has emerged as the leader in AI detection, claims to be able to sniff out generated text with 99.7 percent accuracy. Authors have questioned the reliability of its approach, which itself involves AI.
In September, Jerry Falade, the debut author whose agent reportedly pulled the $2 million book deal, alleged that detection software had misconstrued his Nigerian dialect as evidence he had used AI. Pangram’s software scored the book as 97 percent AI-generated. Falade wrote in a recent essay that it takes months or years to write a book, while AI-detection software takes two minutes to ruin a life’s work. Pangram did not respond to a request for comment.
Others have warned against normalising extreme scrutiny of authors. They claim the labelling schemes have it backward; the onus should be on platforms and publishers to declare when a piece of media is AI-generated, not when it is written by a person. West said authors are being constantly quizzed about the origins of their ideas and work, which is something they need to resist. He added the default should be that a book is human-authored.
A different approach
Approved by the Intellectual Property Office alongside the mark, Books by People’s verification process is different to Pangram’s technology-first approach. It places greater weight on interrogating documents and communication trails for signs of humanity. Dennys said she would not call it AI detection because she realised early on that getting into that race is never going to be sustainable as AI models get better.
In cases where Books by People is uncertain whether a piece of evidence indicates AI use, it claims to err on the side of the author. Dennys said the bedrock of the organisation is to avoid writers being unjustly accused.
Some authors say the important thing is that publishers underscore their commitment to human-authored literature as a nonnegotiable standard for the industry, in whatever way they can. Chiang said it is just as important for publishers to have a no cheating policy as it is for universities, even if there is no perfect method for enforcement. He called it a statement of values.
What it means
This certification offers a practical way for readers to distinguish between human and machine work without relying on flawed detection software. It shifts the burden of proof away from the author’s reputation and onto a vetted process. For publishers, it provides a clear signal to the market that they are maintaining human standards. For writers, it offers a shield against automated accusations, provided the verification process holds up under scrutiny.
