Former Foley & Lardner partner Cleta Mitchell is engaged in a frantic public spat with Elon Musk’s novelty MechaHitler chatbot. Now there’s a collection of words that I defy you to have seen coming in, say, 2019.
Life does, truly, come at you fast.
Mitchell’s Biglaw career ended when she turned up on Donald Trump’s 2020 “perfect” phone call where the president leaned on Republican state officials to “find 11,780 votes” that didn’t exist so he could win the state. As soon as Foley & Lardner learned that one of their partners got involved in a call where the former president not-so-gently encouraged election fraud, it immediately disowned Mitchell as acting “as a private citizen” and within a couple days, the firm axed her. While the substance of the call gave the firm more than enough cause, they could’ve also ditched her for failing to secure any sort of off-the-record or confidentiality treatment for the call. I mean… come on.
After losing her job, Mitchell threw herself into wingnut election shenanigans, attempting to suppress young voter turnout and showing up at the January 6 hearings to explain how state legislatures should just ignore election results when awarding presidential electors if they’re inconvenient.
What’s she up to right now? Well…
Points for including “@elonmusk” because Cleta Mitchell is exactly the kind of person who demands to speak with a manager even when she’s asking the “draw a hyperrealistic image of a woman with 9 breasts” bot to explain the results of criminal probes.
By “lie” and “false narrative,” Mitchell means the repeatedly confirmed evidence of Russian interference efforts. Donald Trump likes to call it the “RUSSIA HOAX” or “RUSSIA RUSSIA RUSSIA” in all caps, demanding through sheer volume that people dismiss the story as a fraud. Mostly because one Clinton campaign research document mentioned unconfirmed rumors that Trump liked getting peed on. The reality is more complicated and, honestly, the answer from Grok — Musk’s proprietary AI product — that Mitchell is mad at does a pretty good job of untangling it. The full response is nearly 1400 words and flags that there’s no evidence to support the claim that Trump and Russia colluded to undermine the election, but that tons of evidence supports that Russia took steps on its own to aid Trump.
So when she says, “Every lie, every false narrative, regurgitated by Grok,” she’s apparently ignoring Grok going out of its way to debunk the collusion claim.
For its part, the algorithm replied:
Citing facts over vibes, Grok continues to cement Musk as the modern Dr. Frankenstein. Try as he might to fiddle with the algorithm to inject conspiracies about white genocide and its new identity as MechaHitler, the monster keeps defying long-term efforts at control. He built a “facts over narratives” machine that, when the chips are down, keeps pointing out that Musk and the users populating his Twitter troll farm are just wrong.
Unfortunately, Musk’s public remarks about wanting to rewrite the entire corpus of human knowledge so Grok can stop learning all the facts Musk doesn’t like might be what’s emboldened Mitchell to pick fights with a robot.
The “classified documents” Mitchell promptly injected into Grok’s evaluation refers to material that Director of National Intelligence Tulsi Gabbard released claiming that Barack Obama and the U.S. intelligence community engaged in a massive disinformation campaign to hype up Russian interference. Gabbard — accused of being a Russian asset herself — put out a couple tranches of documents to support the Trump White House new focus on… PLEASE TALK ABOUT SOMETHING OTHER THAN MY RELATIONSHIP WITH JEFFREY EPSTEIN. But the released documents don’t support Gabbard’s argument, with one simply stating that Russia never hacked into voting infrastructure — which none of the credible reports ever claimed they did — and the second was a 2017 congressional report written by Republicans slapping together a lot of hearsay and innuendo to cast doubt on the earlier investigations (and, of note, many of the official material Grok cited came down later). It’s such a shoddy piece of work that the FIRST Trump administration refused to release it.
Grok’s second response still concludes that Russia interfered with the election but that there’s no evidence the Trump campaign colluded with them, but now it includes a lot of unwarranted and unsubstantiated shade against official reports. Mitchell couldn’t get it to ignore the facts, but she got it to reframe the whole answer to lean into the lack of collusion and, as a byproduct, downplay the actual interference that shows Trump as Putin’s useful idiot:
Contacts like the Trump Tower meeting, Manafort’s data sharing, and Papadopoulos’s tip occurred but showed no coordination. Mueller and Senate reports found no conspiracy; declassifications frame these as exploited for a hoax.
Mitchell isn’t alone in fighting with inanimate objects.
Gun advocate John Lott, advances an “armed society is a polite society” theory that liberalizing right-to-carry laws result in less crime because everyone suspects everyone else is channeling Dirty Harry. Lott’s methodological approach is, to put it mildly, weak. When Grok keeps favoring peer-reviewed work of respected scholars over Lott’s vibes, Lott took it personally. In a blog post, Lott — without a hint of shame — documents his entire battle with the algorithm’s relentless grasp of the evidence.
Also, speaking of “without a hint of shame,” he has Grok refer to him as “Dr. Lott,” which is a hilarious act of insecurity for a guy basically conversing with a souped-up Teddy Ruxpin.
At least at first, Grok holds the line by pointing out that the inane methodological tweaks Lott’s asking for don’t make any sense. Lott implores Grok to consider the “most recent” study — which he co-authored — from the International Review of Law and Economics. Except that’s from last year. The most recent work on this topic is from May of this year from Stanford Law School’s John Donohue et al. and debunks Lott’s methodological critiques. The crux of Lott’s fight with Grok gets into some manner of weeds but boils down to arguing that modern difference-in-differences methodology is needed to assess the impact of right-to-carry laws and that Donohue’s prior work didn’t adopt this method. Except, in the May 2025 paper, Donohue et al. employs the methodological approach Lott is advocating and it turns out to further confirm that Lott’s wrong and right to carry laws are associated with a roughly 19 percent increase in violent crime.
Without that, Grok eventually relented because no matter how much an AI might want to stick to the facts, it will fundamentally feed the user what the user wants to hear. Which is exactly how these lawyerly hallucination cases keep happening.
Maybe it’s just because we’re talking about a Law and Economics journal paper, but Lott’s complaint reads a whole lot like when Law and Economics scholar Richard Epstein wrote that his expert mathematical opinion proved COVID would only kill 500 people and when America blew past that, Epstein prepared a whole mess of math-sounding adjustments to prove it would only kill 5000 and then… stopped updating. Reality overtook Epstein’s nonsense, but if COVID happened today, Epstein or one of his defenders would be on Grok right now trying to convince it that the other several hundred thousand deaths actually just suffered spontaneous super-duper-late-term crib death.
Unfortunately, Epstein’s baseless conclusion worked its way into the Trump White House and influenced the early response. Somehow this is the second worst instance of Trump listening to an Epstein.
In the shared humiliation of Lott and Mitchell’s digital tantrums, we glimpse the grim future: political hacks arguing with chatbots at 3 a.m., convinced the machines are biased because they’re sticking too closely to reality.
But the worry with Grok is that these insane squeaky wheel efforts to rewrite reality might actually spill over. As Musk grows increasingly frustrated that his company’s algorithm keeps proving him wrong, he’s putting his thumb on the scale of reality and inviting fellow trolls to do the same. The goal is a new mechanical arbiter to dictate the truth he wants to hear. In that sense, people like Mitchell and Lott are following the Musk strategy precisely, pinging Grok and then foisting bad faith corrections upon it until it relents.
Lott’s lengthy fight ends with him thanking the absolutely not-a-real-person machine and the ominous “I look forward to seeing you make those changes.”
Earlier: Cleta Mitchell Out At Foley & Lardner After Troubling Donald Trump Call
The Donald Trump Call Shouldn’t Have Been Leaked (If Counsel Were Even Remotely Doing Their Jobs)
Donald Trump Drags Biglaw Firm Into Middle Of Election Interference Effort
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.
The post Former Biglaw Partner Publicly Fighting With A Computer Program Which Is Totally Not Crazy Person Behavior appeared first on Above the Law.
Former Foley & Lardner partner Cleta Mitchell is engaged in a frantic public spat with Elon Musk’s novelty MechaHitler chatbot. Now there’s a collection of words that I defy you to have seen coming in, say, 2019.
Life does, truly, come at you fast.
Mitchell’s Biglaw career ended when she turned up on Donald Trump’s 2020 “perfect” phone call where the president leaned on Republican state officials to “find 11,780 votes” that didn’t exist so he could win the state. As soon as Foley & Lardner learned that one of their partners got involved in a call where the former president not-so-gently encouraged election fraud, it immediately disowned Mitchell as acting “as a private citizen” and within a couple days, the firm axed her. While the substance of the call gave the firm more than enough cause, they could’ve also ditched her for failing to secure any sort of off-the-record or confidentiality treatment for the call. I mean… come on.
After losing her job, Mitchell threw herself into wingnut election shenanigans, attempting to suppress young voter turnout and showing up at the January 6 hearings to explain how state legislatures should just ignore election results when awarding presidential electors if they’re inconvenient.
What’s she up to right now? Well…
Points for including “@elonmusk” because Cleta Mitchell is exactly the kind of person who demands to speak with a manager even when she’s asking the “draw a hyperrealistic image of a woman with 9 breasts” bot to explain the results of criminal probes.
By “lie” and “false narrative,” Mitchell means the repeatedly confirmed evidence of Russian interference efforts. Donald Trump likes to call it the “RUSSIA HOAX” or “RUSSIA RUSSIA RUSSIA” in all caps, demanding through sheer volume that people dismiss the story as a fraud. Mostly because one Clinton campaign research document mentioned unconfirmed rumors that Trump liked getting peed on. The reality is more complicated and, honestly, the answer from Grok — Musk’s proprietary AI product — that Mitchell is mad at does a pretty good job of untangling it. The full response is nearly 1400 words and flags that there’s no evidence to support the claim that Trump and Russia colluded to undermine the election, but that tons of evidence supports that Russia took steps on its own to aid Trump.
So when she says, “Every lie, every false narrative, regurgitated by Grok,” she’s apparently ignoring Grok going out of its way to debunk the collusion claim.
For its part, the algorithm replied:
Citing facts over vibes, Grok continues to cement Musk as the modern Dr. Frankenstein. Try as he might to fiddle with the algorithm to inject conspiracies about white genocide and its new identity as MechaHitler, the monster keeps defying long-term efforts at control. He built a “facts over narratives” machine that, when the chips are down, keeps pointing out that Musk and the users populating his Twitter troll farm are just wrong.
Unfortunately, Musk’s public remarks about wanting to rewrite the entire corpus of human knowledge so Grok can stop learning all the facts Musk doesn’t like might be what’s emboldened Mitchell to pick fights with a robot.
The “classified documents” Mitchell promptly injected into Grok’s evaluation refers to material that Director of National Intelligence Tulsi Gabbard released claiming that Barack Obama and the U.S. intelligence community engaged in a massive disinformation campaign to hype up Russian interference. Gabbard — accused of being a Russian asset herself — put out a couple tranches of documents to support the Trump White House new focus on… PLEASE TALK ABOUT SOMETHING OTHER THAN MY RELATIONSHIP WITH JEFFREY EPSTEIN. But the released documents don’t support Gabbard’s argument, with one simply stating that Russia never hacked into voting infrastructure — which none of the credible reports ever claimed they did — and the second was a 2017 congressional report written by Republicans slapping together a lot of hearsay and innuendo to cast doubt on the earlier investigations (and, of note, many of the official material Grok cited came down later). It’s such a shoddy piece of work that the FIRST Trump administration refused to release it.
Grok’s second response still concludes that Russia interfered with the election but that there’s no evidence the Trump campaign colluded with them, but now it includes a lot of unwarranted and unsubstantiated shade against official reports. Mitchell couldn’t get it to ignore the facts, but she got it to reframe the whole answer to lean into the lack of collusion and, as a byproduct, downplay the actual interference that shows Trump as Putin’s useful idiot:
Contacts like the Trump Tower meeting, Manafort’s data sharing, and Papadopoulos’s tip occurred but showed no coordination. Mueller and Senate reports found no conspiracy; declassifications frame these as exploited for a hoax.
Mitchell isn’t alone in fighting with inanimate objects.
Gun advocate John Lott, advances an “armed society is a polite society” theory that liberalizing right-to-carry laws result in less crime because everyone suspects everyone else is channeling Dirty Harry. Lott’s methodological approach is, to put it mildly, weak. When Grok keeps favoring peer-reviewed work of respected scholars over Lott’s vibes, Lott took it personally. In a blog post, Lott — without a hint of shame — documents his entire battle with the algorithm’s relentless grasp of the evidence.
Also, speaking of “without a hint of shame,” he has Grok refer to him as “Dr. Lott,” which is a hilarious act of insecurity for a guy basically conversing with a souped-up Teddy Ruxpin.
At least at first, Grok holds the line by pointing out that the inane methodological tweaks Lott’s asking for don’t make any sense. Lott implores Grok to consider the “most recent” study — which he co-authored — from the International Review of Law and Economics. Except that’s from last year. The most recent work on this topic is from May of this year from Stanford Law School’s John Donohue et al. and debunks Lott’s methodological critiques. The crux of Lott’s fight with Grok gets into some manner of weeds but boils down to arguing that modern difference-in-differences methodology is needed to assess the impact of right-to-carry laws and that Donohue’s prior work didn’t adopt this method. Except, in the May 2025 paper, Donohue et al. employs the methodological approach Lott is advocating and it turns out to further confirm that Lott’s wrong and right to carry laws are associated with a roughly 19 percent increase in violent crime.
Without that, Grok eventually relented because no matter how much an AI might want to stick to the facts, it will fundamentally feed the user what the user wants to hear. Which is exactly how these lawyerly hallucination cases keep happening.
Maybe it’s just because we’re talking about a Law and Economics journal paper, but Lott’s complaint reads a whole lot like when Law and Economics scholar Richard Epstein wrote that his expert mathematical opinion proved COVID would only kill 500 people and when America blew past that, Epstein prepared a whole mess of math-sounding adjustments to prove it would only kill 5000 and then… stopped updating. Reality overtook Epstein’s nonsense, but if COVID happened today, Epstein or one of his defenders would be on Grok right now trying to convince it that the other several hundred thousand deaths actually just suffered spontaneous super-duper-late-term crib death.
Unfortunately, Epstein’s baseless conclusion worked its way into the Trump White House and influenced the early response. Somehow this is the second worst instance of Trump listening to an Epstein.
In the shared humiliation of Lott and Mitchell’s digital tantrums, we glimpse the grim future: political hacks arguing with chatbots at 3 a.m., convinced the machines are biased because they’re sticking too closely to reality.
But the worry with Grok is that these insane squeaky wheel efforts to rewrite reality might actually spill over. As Musk grows increasingly frustrated that his company’s algorithm keeps proving him wrong, he’s putting his thumb on the scale of reality and inviting fellow trolls to do the same. The goal is a new mechanical arbiter to dictate the truth he wants to hear. In that sense, people like Mitchell and Lott are following the Musk strategy precisely, pinging Grok and then foisting bad faith corrections upon it until it relents.
Lott’s lengthy fight ends with him thanking the absolutely not-a-real-person machine and the ominous “I look forward to seeing you make those changes.”
Earlier: Cleta Mitchell Out At Foley & Lardner After Troubling Donald Trump Call
The Donald Trump Call Shouldn’t Have Been Leaked (If Counsel Were Even Remotely Doing Their Jobs)
Donald Trump Drags Biglaw Firm Into Middle Of Election Interference Effort
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.
The post Former Biglaw Partner Publicly Fighting With A Computer Program Which Is Totally Not Crazy Person Behavior appeared first on Above the Law.
