About a month ago, I wrote Dead Audience Theory and put forward a somewhat unsettling notion: When we believe we’re surrounded by AI slop, we stop engaging with the new and unfamiliar and start behaving like paranoid misanthropes. Over time, authentic human interaction disappears not because AI has replaced every human, but because we no longer trust that anyone is human in the first place.
Since then, the universe decided to stress-test my thesis with some 100% certified gen-u-wine AI-detecting divining rods, Substack integrated Pangram into its website, and the publishing world started cannibalizing its young.
It’s beginning to feel like we’re in that phase of every zombie apocalypse story where the military starts killing everyone, infected and uninfected alike, in a desperate attempt to stop the spread. That never works out. We’re just going to end up with a handful of survivors navigating a world filled with the undead moaning, “I got you, fam,” and injecting little comments about Gremlins and raccoons.
So I thought it’d be best to write a follow-up.
Originally, this was going to be me pointing at our fraying collective sanity and smugly saying, “I told you so.” Instead, somewhere along the way, I had a nightmare revelation:
We’ve spent years teaching ourselves not to trust anything we can’t prove is human. Now that paranoia has reached publishing’s gatekeepers, trusting that an unknown writer actually wrote the manuscript in front of them could become a multimillion-dollar mistake. Eventually, they may decide the safest solution is to stop opening the gate at all.
Since writing Dead Audience Theory, the publishing industry has provided a few increasingly grim case studies to throw onto the book-burning bonfire, making Jamir Nazir, a marginalized author who won a short-story contest only to get tarred and feathered by Pangram and the internet, look almost quaint.
Hachette pulled Mia Ballard’s Shy Girl after allegations that parts of the novel were AI-generated. Ballard denied using AI to write the book herself, saying any AI-produced material might have come from a freelance editor she hired.
Jerry Falade made waves in the literary world with his debut crime novel, Call Me, I’ll Hide the Body, which prompted a fourteen-way bidding war and reportedly earned him an offer of over $2 million before questions about AI use derailed his career almost overnight.
This second case is especially telling.
Fourteen publishers read a manuscript from a complete unknown and decided they wanted it. FOURTEEN.
Even if Call Me, I’ll Hide the Body had been entirely written by ChatGPT, there’s no coherent way to call the finished product “slop.” Fourteen professional publishing houses independently looked at it and decided there was money sitting on those pages. They were probably fan-casting the inevitable film adaptation in the break room.
We should be looking at Jerry Falade, a 20-year-old Nigerian PhD candidate who until very recently looked like publishing’s next breakout author, and be impressed, if not a little resentful, at his sudden success. Again, fourteen publishers were fighting over him. Meanwhile, I’m still getting rejection emails from a manuscript I submitted over a year ago.
The thing I can’t wrap my head around is how nothing about the story itself changed. The words didn’t suddenly get worse once the accusations of AI surfaced. The characters didn’t become less compelling. The prose didn’t spontaneously sprout six fingers. They were still holding the same manuscript. An anonymous editor who bid on the novel told The Bookseller that they didn’t detect any AI while reading it, “and neither did any of [their] sales, marketing or publicity colleagues who also read alongside [them].”
Yet the mere suggestion that AI might have touched Falade’s book was enough for agents and publishers to go from turning the page and seeing a potential bestseller to finding themselves holding a pipe bomb.
At first, I thought about all this like a reader. The thing about text that may or may not be AI-generated, what I’ve come to think of as “mystery meat,” is that once the AI accusation is out there, the work can no longer be judged by something as quaint as its quality. AI slop can’t be good. It can only be considered trash or fraud and then immediately discarded. Simply reading something that might be AI feels like being duped, and the more you happen to enjoy it, the bigger sucker you are.
There’s actually some research backing this up. In a recent Villanova study, 1,498 people read either human-written or ChatGPT-generated short stories without being told their real origins. On pure quality, the AI stories actually scored higher overall. But both human and AI stories received lower ratings when participants were told ChatGPT wrote them.
In two follow-up experiments involving another 905 people, readers were also found to be as accurate as a coin flip at telling the human stories apart from the AI ones.
An even larger study published this year found essentially the same phenomenon across 27,491 participants: evaluations of creative writing consistently fell when people believed AI had written or even helped write it. The researchers call it an “AI disclosure penalty”
In other words, the words themselves weren’t determining their judgment. Their belief about where the words came from was.
Which brings us right back to Falade. His former agent, who dropped him after questions about AI use arose, told The Bookseller, “The book was stunningly good; everyone fell in love with it.” I can’t help but notice the past tense there.
By Falade’s own account, he used a locally hosted LLM for research purposes but did not use AI to write the novel. Sure, a locally hosted model doesn’t leave much of a paper trail, which understandably raises an eyebrow, but the novel is called Call Me, I’ll Hide the Body. It’s probably safe to assume that asking Claude too many research questions on that particular subject is a great way for a Nigerian immigrant to get a visit from Trump’s FBI.
I’m inclined to believe Falade’s explanation, and I certainly wouldn’t consider that kind of AI use disqualifying. The fact that publishing may not see such distinctions as cleanly is particularly terrifying for me because I’m using ChatGPT as a research assistant to write a novella right now.
AI is an amazing tool for writing research. It isn’t penning the novella for me. It’s doing most of the Googling. The story takes place in 1930s Chicago, and ChatGPT can quickly answer extremely specific historical questions, then provide sources so I can verify them. Better yet, it can read the rough draft with that context in mind and point out things like, “Hey, Chicago has greystones, not brownstones,” or, “That random street you pulled off Google Maps didn’t exist in 1932. Use this one instead.”
Initially, I couldn’t see why that kind of AI use would be such a problem for publishers, and the abruptness of what happened to Falade didn’t sit right with me. Surely one of those fourteen publishers would look at the situation and recognize that they were still holding the same manuscript they had fought over. If anything, you’d expect someone to slap a “damaged goods” sticker on it, lop a couple of zeros off the offer and see if Falade would take the deal.
Except the words weren’t the thing that changed. The perceived provenance of the words did.
According to Deadline, after rescinding its submission, Hodgman Literary sent what amounted to a career-destroying burn notice to international publishers: “[The author] produced an incredible manuscript that dazzled us and the top publishing professionals around the world. Unfortunately, we are no longer able to authenticate how the manuscript fully evolved from origin to completion.”
Read that last sentence again. If you’re a writer and you’re not shivering a little, you should be.
“We are no longer able to authenticate how the manuscript fully evolved from origin to completion.”
Can you prove to me that you didn’t use AI in your writing? Can anyone? Taking you at your word simply isn’t good enough, especially if I’m the publisher. The stakes are too high.
Under current U.S. Copyright Office policy, human-authored expression is copyrightable under current law, while purely AI-generated expression is not. A work that includes some AI material may still receive copyright protection, but the protection only covers the portions created by a human, depending on how much creative control the person who created it had.
That distinction sounds perfectly reasonable, until you’re a publisher staring at a manuscript and you have no idea where one ends and the other begins. Tisha Morris, an entertainment attorney who specializes in publishing law, told The Post, “Publishers need confidence that the author can grant the rights promised in the publishing agreement, that the work is eligible for copyright protection, and that it does not contain unauthorized material generated from third-party sources.
So much of this feels like a giant “no duh” moment once you step back and look at how we got here. Silicon Valley gobbled up copyrighted work by the terabyte to train these machines. We shouldn’t be particularly surprised that the legal status of whatever the machine churns out ends up in a murky gray swamp.
If Falade is telling the truth and only used a local LLM for research, none of this should matter to the copyrightability of his novel. But the publishers weren’t in the room while he wrote it. They don’t actually know him. All they have to go on is the two-million dollars worth of manuscript-shaped mystery meat in front of them. The second a question arose about AI, the publisher wasn’t just buying a novel. They were buying uncertainty. Uncertainty over copyright protection. Uncertainty over whether some third party might bring an infringement claim. Uncertainty over the foreign rights, film rights, audiobook rights and translated editions.
And, oh yeah, there’s also the possibility that the entire literary world finds out the publisher paid millions for AI slop. You can’t forget that.
What’s crazy is that the publishers weren’t even the ones who pulled the plug. Falade’s own agent did. Marc Gerald was sitting on a fourteen-way auction, a reported offer north of two million dollars, and what could have been the kind of deal an agent talks about for the rest of his career. But once he could no longer personally vouch for how that manuscript had been created, the safer professional decision was apparently to set fire to his golden ticket.
And I cannot stress this part enough: no one in this whole debacle has publicly produced definitive proof that Call Me, I’ll Hide the Body contains AI-generated prose.
If anything, the way they’re talking about the whole ordeal sounds a lot like somebody consciously trying not to get sued for defamation after they did a whole lotta defaming. They are “no longer able to authenticate” how it was created and can “no longer substantiate” his assurances.
The point I keep coming back to is that what turns text into mystery meat isn’t simply that AI may have been used somewhere in its creation. It’s that we don’t know how AI was used, if it was used at all, and will probably never know.
Everything we read these days is mystery meat on some level. If we don’t know the author personally and can’t verify the chain of custody, we’re just staring at words on a screen and making assumptions. What’s so off-putting about Hodgman Literary’s line, “We are no longer able to authenticate how the manuscript fully evolved from origin to completion,” is that it can easily be applied to every single manuscript submitted since 2019.
So much human creativity is already suffering because we keep looking at the unknown and assuming it’s probably fraudulent simply because AI existed at the time of its creation. Now it increasingly looks like publishers may have a financial and legal incentive to do the same thing. Mystery meat may soon become impossible to publish.
I don’t know what standard writers are being held to, and I’m not convinced the publishing world knows either. If using any form of AI in your writing is completely disqualifying, then that’s a far bigger problem. How do you disprove a negative like that?
Did I just blacklist my 1930s novella by admitting I used ChatGPT for research? What about my future manuscripts? Between Dead Audience Theory and this essay, I’ve clearly established a disturbing pattern of AI use.
What should concern us is everything we don’t see because it didn’t make the news.
How many book deals have quietly died because somebody got a whiff of AI? What Pangram percentage will kill a novel in the crib? Do publishers understand that the model is fallible?
How deeply has this fear penetrated publishing? Are acquisition editors checking old blog posts and social media accounts when provenance questions arise, and being scared off by the slightest bit of evidence that an author touched ChatGPT?
Will writers be asked to document their process? To what extent? To truly “authenticate how the manuscript fully evolved from origin to completion,” agents and publishers would have to see the author’s unpublished drafts, revision history, hell, even internet history. You’d have to scrutinize the entire writing process like it’s Epstein’s prison cell and demand a complete, unedited video recording of the author typing on their computer. God forbid a minute of footage is missing. That’s just enough time for Jerry Falade to open a browser in incognito mode and ask ChatGPT a question about rigor mortis. Now everyone is questioning the authenticity of the suicide note his main character left behind.
There’s already widespread anxiety among writers that AI may increasingly be used to prune the slush pile. Is Pangram going to be yet another filter a manuscript has to get past before human eyes spend five minutes scanning your work? Are agents going to quietly reject anything that trips a detector rather than risk becoming the next person quoted in an embarrassing AI scandal? Faced with hundreds or thousands of submissions, why spend hours investigating whether a 13% AI score on Pangram is a false positive when you can simply move on to the next manuscript?
To be clear, I don’t doubt that most of what gets rejected will be garbage. An impressive amount of it will even be legitimate AI slop.
But the next Fight Club might be sitting in there too.
When I take a step back, I have to ask: What the hell are we even doing? Why are we demanding that writing, a medium already seen as archaic by generations growing up with chronic TikTok brainrot, remain some pure, organic artisanal craft? It’s not a competitive sport. We’re just telling a freakin’ story.
Jerry Falade should be able to ask an AI to look up the effects of decomposition and bloating on a corpse without planting the seeds of his downfall. Authors should be able to use AI without treating any contact with it as disqualifying, as long as they’re honest about what it actually contributed. Prohibition never works. It only turns normal people into liars and frauds, which makes the actual liars and frauds feel right at home.
AI use in writing needs to be graded on a spectrum, not a binary. AI disclosure needs to become a surmountable penalty, not an immediate disqualification. Basically, we need to get to a place where AI-assisted work can be judged on its quality, because we’re never going to stop having mystery meat to contend with. That doesn’t mean we can’t still whip out the Grade-F stamp for the fully generated slop. It just means we have to consider using grades A-D for mystery meat.
The rules around copyrightability and AI-generated material also desperately need more clarity, not because machine slop deserves some sacred legal protection, but because genuine human work is increasingly getting caught in the ambiguity. The second Random House thinks it may have to go through a novel with a fine-toothed comb to figure out which sentences are protected, which aren’t, and whether some copyright troll could plausibly raise a stink, is the second they get the ick and don’t publish that novel.
They simply can’t pass on every unknown author peddling mystery meat, because that will just end with them never publishing any new authors—
—Actually, I take that back. They can do just that.
They basically already are.
For every Jerry Falade, a hundred authors are wondering how the hell he even got Marc Gerald to read page one. I’ve submitted two manuscripts now to countless agents and only ever gotten one email that wasn’t a form letter. Even that one was just two blunt sentences someone had bothered to type themselves. It basically boiled down to, “This is sci-fi and too long. Publishers won’t touch it, so agents aren’t going to look at it. Sorry.”
I stopped wasting my time after that.
Sigh. And that’s what self-publishing is for.
Hopefully Falade will publish the book himself and make everybody look stupid. That doesn’t solve the problem. It just means the gate stayed closed and he had to go around it. In that regard, I wish Falade the very best.
I know I, for one, am excited to read Call Me, I’ll Hide the Body when it comes out.
However it ends up on my Kindle.
Thanks for reading. Keep an eye out for the next chapter in this whole mess: Total Pangramonium, releasing later this week.