Kash Patel’s massive lawsuit against The Atlantic asserts dubious defamation claims, but could it also be… AI slop? The $250 million complaint filed in D.C. federal court — a necessity to avoid the anti-SLAPP laws that would almost certainly make this a financial albatross for Patel down the road — largely underwhelmed. Irrelevant preening, misspellings, and a complete disregard for the looming “actual malice” roadblock made the complaint look less like a serious defamation claim and more a desperate performative plea to convince Donald Trump that Patel is “a fighter” who doesn’t deserve to be kicked to the curb like so many other scandal-plagued administration officials.
Big Lie lawyer Jesse Binnall — who launched this case by promptly publicizing allegations that The Atlantic did NOT actually print, exposing his own client to claims they contend are defamatory — managed to get the lawsuit filed first thing on Monday. While we focused on the flimsy legal claims and embarrassing strategic choices, Caroline Stout locked in on some curious phrasing:
This rhetorical device, which I call “Digiorno Parallelism” as in “it’s not delivery; it’s Digiorno,” ranks among the more infamous signs of AI. It’s not fair to blame AI every time you run across this construction — just like not every em-dash is computer-generated — but it doesn’t inspire a lot of confidence in human authorship.
You might assume — as I did — that AI is not dumb enough to use words like “feable” or “politices,” two prominent misspellings in the complaint, and presume that indicates human writing. While it might prove to be human slop, those errors are, as I discovered, the rare sorts of misspellings AI can make. Most humans would misspell “feeble” by leaving out an “e,” but LLMs are more prone to make phonetic spelling errors, such as imagining a “fea-” opening. Likewise, “politices” is the sort of token sequence confusion an LLM can fall prey to, especially in a complaint that bounces between politics and policies.
That said, humans definitely wrote some of this complaint. Most AI products understand the legal process well enough not to litter the text with gratuitous, irrelevant editorializing. There are too many asides like suggesting the reporter wasn’t “a minimally competent journalist,” or other bush league flourishes that AI would be too professional to include unprompted.
Still, Stout’s observation got me looking at the text and finding more and more quirky language. So I decided to run the whole complaint through an AI detector to see what’s up. What does TextGuard have to say:
Now, recently an AI detector decided that Mary Shelley’s Frankenstein was likely AI-generated, so take these findings with the appropriate margarita rim of salt. Complaints are stylized documents, and it’s easy for a detector to confuse their repetition for an LLM.
But even if the lawyers used AI to create a draft complaint… there’s nothing necessarily wrong with that!
It’s not a party foul to use AI tools to generate a complaint under time pressure — even if the time pressure in this case was entirely self-imposed by a client hoping to seize the narrative upper hand. Complaints (should be) relatively mechanistic, making them prime candidates for AI collaboration. Frankly, AI could help an overeager human lawyer from allowing zealous bloviating take over the cold, formal tone plaintiffs should aspire to bring to their complaints. As long as the user doesn’t go back and tell the algorithm to punch up the vitriol.
Purists may cringe at some of the AI-isms — and human editors should be vigilant in weeding them out — but AI will increasingly be a fixture of the lawyer workflow. If Patel’s lawyers weren’t using it, they were sacrificing efficiency. An AI complaint isn’t creating a lot of the mischief we associate with AI. Ideally, it’s derived from a clear timeline and documented facts, just rearranged into a standard format. That’s where AI typically shines, if the user will let it. It’s not like a complaint includes any hallucinated law.
Unless you count whatever Binnall read that made him think he could clear the actual malice hurdle with “[n]umerous Atlantic pieces over the past two years have characterized Director Patel as unqualified, dangerous, corrupt, or mentally unstable.” Because that case very much does not exist.
Earlier: Kash Patel’s $250 Million Defamation Lawsuit Looks Better With Beer Goggles
FBI Director Promises To Pound ‘The Atlantic’ Like A Six Pack On A Tuesday
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.