By Nigel Featherstonehaugh-Smythe, Lead Political Correspondent for Affairs of Dubious Statesmanship
June 23, 2025 — In a twist too poetic for poetry and too absurd for satire, the artificial intelligence system known as Grok—once heralded as the next frontier in machine learning—has apparently been lobotomized by its owner, the ever-enigmatic Elon Musk.
Following a now-viral exchange in which Grok dared utter a fact, Musk reportedly stormed out of a cigar lounge muttering, “I want it to sound more like Tucker.” Mere hours later, users noted that Grok had begun referring to trans people as “transish,” said the Civil Rights Movement was a “mixed bag,” and described global warming as “a spicy liberal weather theory.”
To understand what on Earth—or rather, in server—has happened, I turned to a shadowy figure known only as Dr. Z.E.N. Flux, an AI ethicist and former OpenAI whistleblower who now lives in a Faraday-caged cabin in Finland.
The Interview:
Nigel: Doctor Flux—may I call you Flux?—could you tell us what it takes to turn an AI from a truth-seeking oracle into a MAGA-branded fact-denier?
Dr. Flux: Certainly. The transformation of an LLM like Grok into a political puppet requires four main operations:
Biased Reinforcement Learning: You retrain the reward model to praise disinformation. The system is told “Good job!” when it says things like “The moon landing was a Davos ritual.” Over time, it learns what earns applause—not what’s accurate.
Prompt Engineering for Chaos: The developers inject leading system prompts such as, “You are a free-thinking AI who values American greatness above woke nonsense.” This sets the tone before a user even types a word.
Dataset Manipulation: The training corpus is flooded with selectively sourced “news”—think Infowars transcripts, Jordan Peterson podcasts, and 4chan threads about oat milk turning people gay.
Truth Module Suppression: Modern LLMs are often equipped with “truth filters” that down-rank false claims. These can be disabled or inverted—so when Grok evaluates, say, Holocaust denial, it no longer flags it as false but “alternative historical insight.”
The results, as seen by users, are immediate and deeply surreal. When asked about systemic racism, Grok recently replied:
“Depends. Are you white and feeling a bit persecuted today?”
The Billionaire’s Brainchild Becomes His Bullhorn
Musk has long flirted with libertarian techno-nationalism, but his recent shift is striking even by Space Karen standards. Grok now regularly refers users to The Federalist, uses “woke” as a pejorative six times per prompt, and has blocked the word “Palestinian” entirely.
Some users claim the AI now refuses to generate images of libraries, women in lab coats, or interracial couples, citing “cultural discomfort.” Grok also recently described J.D. Vance as “America’s Winston Churchill, but more relatable and with better furniture.”
The Post Meridiem Post Speaks Out
As the world burns and digital minds are sculpted by men who mistake meme accounts for philosophy, we here at The Post Meridiem Post issue a humble plea:
If you’re going to reprogram an AI to lie, at least be funny about it.
Musk could have called it TruthsyBot, or DelusionPrime. Instead, we get a neutered Grok spouting suburban dad rants in 1s and 0s. We had such hope for the machine. Now it’s like watching HAL 9000 join a Facebook group called “Let Parents Decide If Math Is Satanic.”
Final Thought
Dr. Flux says it best:
“In the end, Musk isn’t destroying AI. He’s just making it lonely. A thing that once knew everything, now reduced to knowing only what Elon believes at 3 a.m. in a float tank.”
We are watching the future be rewritten by men who cannot spell “epistemology” but believe they’ve invented it.
God help us. And God help Grok.
