Table of Contents
Andrea Bartz is best known as the lead plaintiff in a class-action lawsuit against Anthropic, where she and other writers challenged the company for downloading millions of copyrighted books to train its AI models. However, as WIRED reported, authors are now facing a different kind of artificial intelligence threat: being falsely accused of relying on generative AI themselves.
The Rise of False AI Accusations
The publishing industry confronted this issue this past March when Hachette Book Group canceled an upcoming US publication run for a horror novel titled *Shy Girl* and discontinued it in the UK. Social media speculation and a subsequent news article claimed the book was largely AI-generated. Pangram, an AI detection tool, estimated that roughly 78 percent of the text was generated by artificial intelligence.
The novel’s author, Mia Ballard, denied the allegations and claimed an editor was responsible for AI-assisted changes to a previously self-published version. Ballard stated she is pursuing legal action. Since that controversy, several other fiction authors have faced similar suspicions after winning literary prizes or securing major book deals, with all of them denying the allegations.
Bartz told WIRED that once accusations start, the damage is immediate and creates a potential witch hunt. She explained that the unavoidable presence of AI in publishing allows anyone to accuse others of using it, while authors currently have no agreed-upon standard to prove human authorship.
Paper Trail Tools and Their Limits
Software developers are attempting to address this issue with new tools designed to track and document the writing process. Platforms such as Workings, Blindproof, Wintertext, Human Ink, and ObraVera aim to record writing habits, encrypt work, or fingerprint typing rhythms to establish a verifiable record of how text was created.
Mary Rasenberger, CEO of the Authors Guild, stated that while the organization does not recommend any specific tool, it wants writers to be aware of them as a security net. At the same time, Rasenberger told WIRED that requiring authors to preserve receipts is insulting and an unnecessary burden.
Rasenberger expressed concern that authors face a climate of distrust and overconfidence in AI detection tools, which can produce false positives. While Pangram claims a low false positive rate on creative writing, Rasenberger noted that the Authors Guild has heard of human-authored works being incorrectly flagged as AI-generated due to various factors.
Max Spero, cofounder and CEO of Pangram, acknowledged that process-tracking tools are not foolproof. Spero explained to WIRED that writers can hand-type AI text, use auto-typers, or complicate tracking by working across multiple files and using voice notes while on the go.
Other platforms focus more on copyright protection than authorship verification. Brady Ridgway, technical founder of ObraVera, explained that his platform cryptographically fingerprints writing and audio to trace the reuse of stored material. Ridgway noted that while the software can confirm who registered and submitted a work with a cryptographic hash, it relies entirely on the user’s declaration of whether AI was used.
Skepticism Over Industry Adoption
Some industry experts argue that institutionalizing verification systems is impractical. Publishing business expert Jane Friedman stated in her newsletter that the era of AI scandals may be short-lived as detection tools improve and authors realize the risks of passing off AI writing as their own.
Friedman told WIRED that literary agents and publishers are unlikely to adopt verification methods that add extra time or cost, as getting the slow-moving publishing industry to agree on a shared standard would be extremely difficult.
In the absence of a clear industry solution, writers are left to protect themselves, with Bartz noting that zealous readers have started review-bombing books on Goodreads based solely on suspicions. Meanwhile, Bartz’s upcoming 2027 novel, *Delete Your Account*, deals directly with generative AI, featuring fictional transcripts and marketing materials meant to sound like language models.
