New tools aim to help authors prove that AI didn't write their text
Following the Shy Girl scandal, in which the detection tool Pangram flagged 78% of the text as AI-generated, developers are offering software such as Workings, Blindproof, Wintertext, Human Ink or ObraVera that tracks the writing process as evidence of human authorship.
Publisher Hachette Book Group cancelled the US release of the horror novel Shy Girl this past March and withdrew it from sale in the UK after speculation appeared on social media and in the Times that the text was largely AI-generated. The detection tool Pangram flagged approximately 78 percent of the text as AI-generated. Author Mia Ballard denied the accusation and says she is considering legal action; similar accusations have also been faced by other authors, including the winner of a short story competition and the author of a crime novel that publishers had been bidding on for millions of dollars.
According to Mary Bartz, a plaintiff in the lawsuit against Anthropic (which last year agreed to a $1.5 billion settlement over the use of pirated books as training data), there is no agreed-upon standard for proving human authorship, which means accusations of using AI can harm an author with no way to defend themselves.
Several new software platforms are responding to this gap — Workings, Blindproof, Wintertext, Human Ink and ObraVera — which, according to their developers, track and document the writing process, in some cases encrypting and fingerprinting the text or writing rhythm using unique codes. The goal is to create proof of authorship that can be used both when accused of using AI and when demonstrating a text's origin if it was misused in training data without attribution. According to CEO Mary Rasenberger, the Authors Guild does not recommend any specific tool, but wants authors to be aware of their existence as a safety net; at the same time, it considers the requirement to keep such "proof" a burden and "offensive."
Industry representatives point to limitations: detection tools like Pangram can produce false positives even on purely human-written text, and tracking tools can be circumvented (e.g. by manually retyping AI-generated text or using an auto-typer). Pangram CEO Max Spero and ObraVera technical founder Brady Ridgway both confirm that none of the approaches is foolproof, and that the real writing process (multiple files, voice notes, collaboration with an editor) is not fully captured by these tools.
Why it matters
A new practical question is emerging for authors and publishers: how to protect against unproven accusations of using AI, which can lead to the cancellation of a release or contract even before any verification takes place. Writing-process-tracking tools offer some form of proof, but according to the experts cited, they are neither reliable nor foolproof, and detection tools such as Pangram can generate false positives even for purely human-created work.
Two audiences, two different impacts
What this means
For individuals
Authors who want proof of the origin of their text may consider tools that track the writing process, but must keep in mind that these are not a hundred percent protection against accusations or against circumvention.
For a business
Publishers face the risk that even unproven suspicion of AI-generated text (spread on social media) can lead to the cancellation of a publication, as happened with Hachette Book Group and the novel Shy Girl; detection tools such as Pangram are not reliable enough for a definitive decision.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.