There’s a new kind of literary scandal sweeping the world of letters—and it’s not just about plagiarism or ghostwriters. It’s about the quiet, creeping betrayal of artificial intelligence, the tool that promised to liberate writers from drudgery but has instead become the unwitting architect of some of the most embarrassing gaffes in publishing history. Steven Rosenbaum, a media entrepreneur and author of The Future of Truth, recently found himself in the eye of this storm after admitting his book contained multiple fake or misattributed quotes—some of which, he now claims, were “fucked up” by ChatGPT itself. But Rosenbaum’s confession isn’t just about one man’s missteps. It’s a symptom of a much larger crisis: the moment AI stopped being a tool and started being a wildcard in the writing process.
The problem isn’t that AI writes poorly—it’s that it writes too well. Just last week, a Nobel-winning novelist faced backlash after a viral post suggested she’d used AI to refine her work, only to later claim she’d been “misunderstood.” Meanwhile, the Commonwealth Foundation scrambled to investigate allegations that Jamir Nazir, the winner of the 2023 Commonwealth Short Story Prize, had outsourced his entire manuscript to a chatbot. By Wednesday, two more prize winners were under scrutiny. The foundation’s initial denial—“none of the winning writers had used AI”—was followed by a humiliating retraction: they were “taking the allegations seriously.”
This isn’t just a literary scandal. It’s a crisis of trust. And the deeper you dig, the more confusing it gets.
How a Chatbot Became the Fall Guy
Rosenbaum’s about-face is telling. Initially, he blamed himself for failing to verify AI-generated quotes—including one supposedly from tech journalist Kara Swisher, which turned out to be pure hallucination. But by the time he spoke to reporters, he’d shifted blame to ChatGPT, calling it “quirky or evil or sneaky.” His frustration is understandable. AI tools like ChatGPT are designed to mimic human thought, but they do so with staggering inconsistency. One moment, they generate prose so polished it fools even seasoned editors; the next, they spit out fictional quotes, fabricated sources and passages so clunky they read like a first-year student’s attempt at academic jargon.
Archyde analyzed The Future of Truth using Pangram, an AI detection tool, and found a 146-word passage that registered as 100% AI-generated. When confronted, Rosenbaum dismissed the tool’s findings, but the damage was done. His book—meant to critique AI’s role in shaping reality—had become a cautionary tale about its own limitations.
“The real villain isn’t the tool,” says Dr. Tim Requarth, a neuroscientist and AI ethics researcher at NYU. “It’s the dependency. Writers are outsourcing the hard work of discovery—research, synthesis, even moral judgment—to systems that don’t understand the difference between truth and fabrication.”
Requarth’s point cuts to the heart of the issue: AI isn’t just replacing words; it’s replacing thought. And that’s where the scandals get dangerous.
The AI Spectrum: Where’s the Line?
The problem isn’t monolithic. There’s a spectrum of AI use in writing, and the scandals expose how blurry the boundaries have become:
- Maximalist AI: Typing a prompt like “Write a haunting short story in the style of Toni Morrison” and slapping your name on the output. (This is the kind of abuse that gets headlines.)
- Minimalist AI: Using chatbots to suggest a better word, verify a fact, or draft an outline. (This is where most professionals operate—and where the ethical gray area lies.)
- Hidden AI: Relying on AI-powered search results, AI-curated sources, or even AI-generated summaries without realizing it. (This is the most insidious form, because it’s invisible.)
The Authors Guild warns of “ethical risks” in all three categories, but their guidelines are vague. The New York Times, for instance, allows freelancers to use AI for “high-level brainstorming” but bans it for newsroom staff—except when they’re encouraged to “experiment” with it as a “powerful tool.” Even The Atlantic, which has its own internal AI guidelines, admits to using chatbots for transcription and “smarter thesaurus” functions.
“The real threat isn’t the overuse of ‘delve’ in academic papers,” Requarth adds. “It’s that we’re outsourcing the process of understanding the world to machines that don’t grasp context, bias, or nuance.”
Consider this: A recent Amazon working paper estimated that over 50% of new books on the platform contain AI-generated text. Yet until recently, most of these books flew under the radar because their prose was quality enough to pass as human. Now, as AI improves, the bar for detection is rising—and so is the stakes.
Who Wins and Who Loses in the AI Writing Arms Race
The economic and cultural ripple effects of this shift are only beginning to surface. Here’s who stands to gain—and who’s already getting burned:
- Winners:
- AI Startups: Companies like OpenAI and Anthropic are seeing surging demand for enterprise-grade writing tools, with subscriptions from media outlets, law firms, and even government agencies.
- Plagiarism Detection Firms: Tools like Turnitin and Copyscape are pivoting to AI detection, charging universities and publishers premium rates to verify originality.
- Self-Published Authors: Indie writers with limited budgets are turning to AI to cut costs, flooding markets with high-volume, low-effort content that undercuts traditional publishers.
- Losers:
- Traditional Publishers: The Association of American Publishers reported a 12% decline in book sales last quarter, with AI-generated titles cannibalizing mid-list authors.
- Academic Integrity: Universities are scrambling to update plagiarism policies. A Chronicle of Higher Education survey found that 68% of professors have caught students using AI to write essays—up from 22% in 2023.
- Literary Reputation: The Commonwealth Short Story Prize scandal has already led to calls for mandatory disclosure of AI use in submissions. If enforced, this could devalue prizes by making them a game of “who can hide their AI best.”
The most alarming trend? AI is no longer just a writing assistant—it’s a research assistant, an editor, and sometimes, a ghostwriter. Tech reporter Alex Heath recently trained an AI to mimic his writing style, using it to generate first drafts. “It’s not cheating,” he told Wired. “It’s collaboration.” But when that collaboration leads to misinformation, who’s accountable?
The Paradox of Progress: Why Better AI Makes the Problem Worse
Here’s the counterintuitive truth: The better AI gets, the harder it is to detect—and the more dangerous it becomes.
In 2023, AI-generated text was easy to spot. It sounded robotic, overused transitions like “delve into” and “pivot toward,” and struggled with basic facts. Today? The best models can mimic specific authors, generate plausible but false citations, and even adapt to tone. A study by arXiv found that 72% of human reviewers failed to detect AI-written legal briefs when they were embedded in real cases.
“The real crisis isn’t that AI writes badly,” says Dr. Kate Crawford, a tech ethics researcher at USC. “It’s that it writes convincingly. And when people trust it too much, they stop thinking critically.”
Crawford points to a 2023 Nature study showing that 40% of academics who used AI for research admitted to not verifying its sources—assuming the chatbot’s answers were correct. That’s not just sloppy writing. It’s systemic risk.
And the problem extends beyond text. AI is now being used to:
- Generate fake academic papers (with ScienceDirect retracting over 1,200 AI-written studies in 2024 alone).
- Create deepfake audio of authors reading their own work (used in 17% of recent book promotions, per Nielsen BookScan).
- Automate book reviews on Amazon, inflating ratings for AI-generated titles while suppressing organic sales.
The result? A feedback loop where trust in AI rises, critical thinking declines, and the line between human and machine blurs to the point of invisibility.
The Way Forward: Can We Fix This?
There’s no easy answer. But the scandals have forced a reckoning—and a few potential solutions are emerging:
- Mandatory Disclosure: Publishers like Penguin Random House are testing AI usage labels on book covers, similar to nutrition labels on food. The idea? Transparency over shame.
- AI Literacy Training: The Education Week reports that 34 states are now requiring AI ethics courses in high schools and universities.
- Dynamic Detection: New tools like GPTZero are using stylometric analysis to detect AI not by pattern-matching, but by behavioral anomalies in writing.
- Industry Bans: The Guild of Book Workers has called for a voluntary ban on AI in prize submissions—though enforcement remains a challenge.
But the most urgent question remains: What do we lose when we outsource the act of creation?
Rosenbaum’s book was meant to explore how AI shapes reality. Instead, it became a victim of the very technology it examined. The irony isn’t lost on him. “I wrote a book about truth,” he told Archyde. “And then the truth wrote itself.”
That’s the real scandal. Not that AI makes mistakes—but that we’re relying on it to define what truth even is.
What You Can Do
The next time you read a book, an article, or even a tweet, ask yourself:
- Could this have been written—or influenced—by AI?
- Does the author’s voice still feel human?
- Are the sources primary, or are they pulled from a chatbot’s “hallucinated” knowledge base?
And if you’re a writer? The choice isn’t between using AI and not using AI. It’s about how you use it—and whether you’re willing to accept the risks. Because in the age of the chatbot, the real question isn’t Can you trust AI? It’s Can you trust yourself to know the difference?
Now, tell me: How far would you go with AI in your own work? (And more importantly—how would you know if you’d crossed the line?)
