The AI-authorship line is blurring

A New York Times essay about losing custody of a child turned into an authorship fight after writer Becky Tuch posted that a passage “reads EXACTLY like AI slop” and said detector tools had flagged it. The essay was Kate Gilgan’s November 2025 “Modern Love” piece, and the argument blew up in March 2026. (futurism.com) (theatlantic.com) Gilgan did not say “I wrote every word alone,” and she also did not say “a chatbot wrote the essay for me.” In an interview with Futurism, she said she used artificial intelligence to organize material, compress information, and help shape a submission aimed at getting into the Times. (futurism.com) That middle ground is the whole problem. “AI-generated” sounds like pressing a button and getting a finished draft, but “AI-assisted” can cover everything from brainstorming a headline to rewriting the bones of a personal story. (futurism.com) (poynter.org) The New York Times already has a rule for this, and the rule is squishy in exactly the place people now care about most. The company says substantial use of generative artificial intelligence should be disclosed to readers, but “substantial” leaves room for argument when a writer uses a chatbot as a structural editor instead of a ghostwriter. (nytco.com) (newsbreak.com) That ambiguity is not limited to one essay. Nieman Journalism Lab reported in February 2026 that freelancers were already offering editors Google Docs version histories as proof they had actually done the writing themselves, which is a sentence that barely made sense in 2022 and now sounds normal. (niemanlab.org) Detection tools are not solving this. Even coverage that amplified the accusation also noted that artificial intelligence detectors are unreliable, which means a suspicious paragraph can start a public trial without producing anything close to courtroom-grade evidence. (futurism.com 1) (futurism.com 2) That leaves editors judging process, not just prose. If a freelance writer uses a chatbot to turn notes, memories, court records, and old drafts into a cleaner narrative, the final piece may still contain true facts and real experience while no longer feeling fully human-made in the old sense. (futurism.com) (onlinejournalismblog.com) The fight around Gilgan’s essay landed just as other news organizations were tightening their own rules. The Globe and Mail wrote on April 4, 2026 that major outlets were revisiting policies after the Times controversy, which shows how fast one disputed essay became an industry-wide warning. (theglobeandmail.com) The practical question now is less “Was this written by AI?” than “Which parts of writing count as writing?” A machine that suggests phrasing, trims repetition, rearranges scenes, and proposes endings can leave a byline intact while quietly changing what readers think that byline guarantees. (futurism.com) (poynter.org) That is why this episode stuck. A personal essay is supposed to feel like the most unmistakably human form on the page, and this one still ended with readers, editors, and the writer herself arguing over where human memory stopped and machine language began. (futurism.com)