What happened: A New York Times "Modern Love" column from November 2025 sparked controversy when writer Becky Tuch posted an excerpt on X, noting it "reads EXACTLY like AI slop." AI researcher Tuhin Chakrabarty ran the piece through detection tool Pangram Labs, which flagged over 60 percent as likely AI-generated. The author, Kate Gilgan, admitted using ChatGPT, Claude, Copilot, Gemini, and Perplexity for "inspiration and guidance" but insisted she didn't paste AI-generated content directly.

Why it matters: This isn't an isolated incident. Research by Jenna Russell at the University of Maryland found AI-detection flags across major US publications including The New York Times, Wall Street Journal, and Washington Post opinion sections. While these papers have AI policies requiring disclosure for "substantial use," the Gilgan case reveals a grey zone where AI functions as a "collaborative editor" rather than outright ghostwriter.

Wider context: The infiltration extends beyond newspapers. Hachette recently cancelled a novel, "Shy Girl," after readers identified AI-generated text. Last spring, the Chicago Sun-Times and Philadelphia Inquirer published a syndicated summer-reading guide featuring nonexistent books—created by a freelancer using ChatGPT. The machines aren't just coming for content farms anymore.

Background: AI detection tools remain imperfect, with false positives and varying results across platforms. But the pattern is clear: prestigious media outlets, once considered protected spaces from AI slop, are showing cracks. The social contract between readers and publications—that attributed opinions reflect genuine human voices—faces an existential challenge when AI output is more homogenous and politically skewed than human writing.

How AI Is Creeping Into The New York Times — The Atlantic

Singularity Soup Take: The irony is delicious—publications that built trust on human judgment are now discovering their own writers can't resist the siren call of algorithmic assistance. When even the Paper of Record needs AI to help people express their feelings, perhaps the problem isn't the tools but the expectation that authentic emotion should arrive on deadline.

Key Takeaways:

• Detection Discrepancies: Five different AI detection tools produced wildly different results on the same column—two flagged 30% as AI-generated, one found nothing, and Pangram claimed 60%+.
• Policy Loopholes: The Times requires disclosure for "substantial use" of generative AI, but Gilgan's case shows how "collaborative editing" can slip through without triggering transparency requirements.
• Research Evidence: A University of Maryland preprint study using Pangram found likely AI use across opinion sections at major US newspapers, though far fewer news stories were flagged.
• Persuasion Risk: Multiple studies show AI output is unusually persuasive at changing political beliefs—a concerning trait when undisclosed machine language infiltrates influential opinion columns.