After thirty years of watching the publishing industry reinvent itself, I can tell you this: AI is not all good or all bad for writers. What matters is whether it supports your work or replaces it, and whether your content was stolen to build it.

AI Detection Tools are Not Perfect

Writing a novel can take months or years. Within two minutes, AI-detection software potentially can ruin an author’s career, even if he/she claims the entire story was human created.

Jerry Falade, the author of Call Me, I’ll’ Hide the Body, knows the anguish of AI-detection tools. Rightly or wrongly, his publisher cancelled a $2 million publishing deal after AI flagged the work as machine-generated. (Source: frontlist.in).

While AI detection tools like Pangram, GPTZero or Turnitin, do catch many unscrupulous authors and purveyors of illegally copied books, the tools are fallible and sometimes biased.

Where is the Joy in a ‘Defamation Machine?’

Falade maintains he himself created the crime novel, including characters, plot and other elements. In a recent essay he posted on Kathleen Schmidt’s Substack.com, he asked, “If creative people have to second-guess every natural instinct because someone might call it an ‘AI pattern,’ then where is the joy in writing?…at what point does proving we are human require us to stop sounding like ourselves?”

In April 2026, Wall Street Journal editor James Taranto called the AI detector a “defamation machine,” after Pangram tagged three of his articles as AI-written. He called its use unfair. Many similar complaints have been publicly reported by writers this year alone.

The Authors Guild Reacts

Falade’s experience among others prompted The Author’s Guild (October 2026)  to craft a common sense approach for publishers and agents on the use of AI detectors and dealing with undisclosed AI use.

While agents and publishers ask writers to be transparent about using AI, The Guild now asks those pros to follow seven guidelines when considering an author’s story. I paraphrase here.

Key guidelines for agents and publishers on the use of AI

    1. Have a clear, written AI policy of part of an author’s contract. Never cancel a contract over AI use unless the author breaches the signed contract.
    2. Don’t use AI to edit beyond spelling and grammar, and don’t add content without the author’s written permission.
    3. When detectors are used in the editorial process, be transparent about their use. Get the author’s permission if sending a manuscript to a third-party detector. Also don’t allow the text to be kept for AI training.
    4. Never rely on a detector alone. Consult with the author if you are concerned.
    5. Use detector results to start a conversation. Give the author a chance to rewrite. Never shame an author publicly.
    6. Some forms of writing , such as non-native English speakers, genre writers and those with a deliberately plain style can raise false flags. The editor or agent should treat those flags as weak evidence.
    7. Don’t require authors to prove they authored their books with tracking software. The Guild does support authors who choose to keep their drafts.

Face to Face With AI

For this piece, I ran an experiment beginning with a jumble of ideas then feeding the raw draft through Claude which suggested it could write the piece in my own voice. I said yes. The result was way off base. I gathered a few ideas from the result and began all over again as my human self.

Later, I  directly asked Claude to “list at least five word patterns AI detection tools flag as AI-written.”

It is important for writers to know the answer, posted here. And it might mean this article could be flagged, though all words for this iteration have been typed by me unless clearly indicated. The concepts and structure have been honed from about fifty notes, ideas and research tracked in Word Press. So here is the AI answer.

Claude AI Tells Me How it Works

AI detectors mostly score how predictable text is rather than checking a word list, but certain patterns reliably raise those scores and are widely flagged by detectors and human readers alike:

    1. Overused “AI vocabulary.” Words like delve, tapestry, testament, realm, pivotal, underscore, multifaceted, intricate, foster, robust, seamless, and navigate the landscape show up far more often in AI output than in human writing.
    2. The “It’s not just X, it’s Y” construction. For example, “This isn’t just a book, it’s a journey.” Its cousins include “More than X, it’s Y” and “X isn’t about A. It’s about B.”
    3. Formulaic openers and closers. Examples: “In today’s fast-paced world…,” “In the ever-evolving landscape of…,” “Whether you’re a seasoned pro or just starting out…,” “In conclusion,” “Ultimately,” and “At the end of the day.”
    4. Stacked transitions. Paragraphs that keep starting with Moreover, Furthermore, Additionally, and It’s worth noting that read as machine-made.
    5. Reflexive groups of three. Examples: “clear, concise, and compelling” or “innovation, collaboration, and growth.” These are especially telling when every list in a piece has exactly three items.
    6. Uniform rhythm (low “burstiness”). Sentences of similar length and structure, paragraph after paragraph, are a strong signal. Human writing mixes short punches with long, winding sentences.
    7. Heavy em-dash use and tidy summary lines. Frequent dashes, plus a neat restating sentence at the end of each paragraph, have become an informal tell.
    8. Vague hedging and empty praise. Examples: “plays a crucial role,” “offers valuable insights,” “a rich and diverse array.”

One caution: these tools produce many false positives. Polished professional prose, formulaic genres like query letters and press releases, and writing by non-native English speakers often get flagged even when entirely human-written. A detector score is a hint, not proof.

I must agree with The Authors Guild policy: “The bottom line is that if an editor suspects undisclosed AI use, they should first contact the author and have a frank conversation. Results from an AI detector should be used as a basis for discussion, not to summarily cancel a contract or publicly shame any author.”

You can read the full guidelines on The Guild’s website.

Writers should do some things to protect themselves against AI suspicions.

  • Keep dated drafts and revision history.
  • Know what counts as AI. Grammarly and similar apps may be ok as AI assistants. You should review every change they suggest. (I myself turn off Grammarly while working. The pop ups have become an irritant, not a help.)
  • If you are fortunate enough to land an agent or publisher don’t be shy about asking for their AI policy and refer to The Guild’s guidance.
  • Honestly disclose any AI use. If a policy conflicts with how you work, submit elsewhere. Don’t hide it.
  • If flagged, ask which tool was used, what score was given and on what pages. Offer your drafts and ask for human read and a conversation, as The Guild recommends.
  • Don’t use “humanizers” to disguise AI text. It puts your reputation at risk. An AI humanizer is a software tool that rewrites text generated by AI so that it sounds more like natural human writing.

The advent of AI has given us unprecedented boundless access to the minds of others. The new technology can be viewed as  the world’s largest library contained inside dark buildings with flickering lights and tangled wires instead of wooden shelves.

To seize control of this vast repository of human creativity, we writers must use AI  responsibly, ethically. If we instruct it to write whole stories without our interaction, claiming  the work as our own, we not only cheat readers, but we also cheat ourselves by giving AI the power to replace us.

Doris Booth, CEO of Authorlink.com

Moral integrity, Ideas, instinct, reasoning, judgement and wisdom belong not to AI, but to us. Keep creating and writing.

Note: One final word of disclosure. For the past two years, my website offers several ways to draft query letters and summaries, using AI responsibly. From the very beginning, we have maintained that an experienced  human editor must always be the thought leader in the writing process. We encourage clients to thoroughly check the AI reports for accuracy and relevance.