Artificial intelligence detection tools are increasingly shaping how people write, with authors reportedly changing their vocabulary, punctuation, and sentence structure to avoid being falsely accused of using AI. The growing concern is that writers are now modifying their natural writing styles to make their work appear more human to detection software.
According to a Bloomberg report, several writers have started avoiding words such as "delve" and punctuation marks such as em dashes, which are commonly associated with AI-generated content. Some are even deliberately introducing mistakes to avoid being flagged by detection tools.
One case highlighted in the report involves writer Mia Ballard. Hachette Book Group reportedly dropped plans to publish her book, 'Shy Girl', after AI detection software Pangram gave it a 78.3 percent AI-written score.
Ballard denied the allegations, claiming that the person who edited her book had used AI for refinement without her knowledge. She later self-published a novella, 'Poly', which she said was written by hand. However, Pangram reportedly identified its first chapter as 100 percent AI-written. Her 2024 book, 'Sugar', also received an AI-written score of 83 percent.
In another novel, 'American Teen Princess', Ballard reportedly reduced poetic language to make the writing sound more human. Pangram subsequently gave it a 37 percent score, identifying it as a combination of human and AI writing.
The report also highlights Nigerian author Jerry Falade, who reportedly lost a multi-million-dollar book deal after being accused of using AI. He has advised writers to avoid what are commonly described as AI 'tells'.
Research from Germany's Max Planck Institute for Human Development also found that the use of the word ‘delve’ increased in human language following ChatGPT's release. Em dashes have similarly become a concern among writers.
Also Read: Google Expands Chrome Security With New Website Alerts and AI Scam Detection
The issue extends beyond authors, with multiple publications reportedly firing journalists accused of using AI to write articles. Some writers now use AI tools to humanize their content, while others change their natural writing habits to avoid suspicion.
However, questions remain about how detection tools can become more reliable and prevent false accusations against human writers.