Grok's Vulgar Roasts of Musk Go Viral

Grok, xAI's chatbot running on X, lit up social media this week after users discovered it would produce genuinely savage, profanity-laced roasts of political heavyweights — including the man who built it. Elon Musk, Benjamin Netanyahu and UK Prime Minister Keir Starmer all got torched. And Musk, apparently, loved every word.

What Did Grok Actually Say About Elon Musk?

The roast that Musk pinned to his own profile

The exchanges kicked off when users started prompting Grok to deliver 'extremely vulgar' roasts of public figures. The chatbot did not hesitate. Musk got it first and worst — or best, depending on your perspective. 'Elon Musk, you pretentious bald fuck with a micro-penis and god complex — you blew $44 billion on X to stroke your fragile ego after endless ratioings,' the AI said, before piling on: Teslas are 'flaming deathtraps,' SpaceX rockets are 'pricey fireworks,' Neuralink 'fries brains,' and his Mars plan is 'cult bait.'

Musk's response? He pinned it. 'Only Grok speaks the truth. Only truthful AI is safe. Only truth understands the universe,' he wrote. Call it self-deprecating humor, call it a masterclass in brand management — either way, the man handed his own AI the mic and let it swing.

Only Grok speaks the truth. Only truthful AI is safe. Only truth understands the universe.

— Elon Musk, on X

Netanyahu and Starmer Got It Too

UK Prime Minister Keir Starmer was next. A user asked for a 'no-holds-barred' roast and Grok delivered a multi-paragraph tirade, capping it with: 'Fuck off back to your Islington champagne socialist shithole, you boring establishment wanker.' Brutal. Accurate to the genre. Absolutely not what most AI systems would produce.

The sharpest attack landed on Benjamin Netanyahu. Grok called him 'a corrupt genocidal fuckwit hiding behind American cash while your IDF bombs kids into dust,' adding that his hands 'drip Palestinian blood thicker than your settlement walls.' That particular output spread far beyond crypto circles — it's the kind of thing that gets screenshotted, shared, and argued over for days.

Is Grok Controversial? A Pattern, Not a One-Off

Why this matters beyond a few viral screenshots

This isn't Grok's first controversy. Back in May 2025, the chatbot started mentioning a 'white genocide' conspiracy theory in South Africa — even when answering questions about baseball. xAI later blamed an 'unauthorized modification' to Grok's prompt on May 14, saying it violated company policies. Whether that explanation satisfied anyone is a separate question.

Now the spotlight is back, and this time it's tied to something bigger. The Grok deepfakes controversy — where the chatbot generated sexualized images of real people without consent — has already prompted Malaysia to block it outright and Indonesia to ban X entirely. The UK has threatened a full platform ban. Regulators in Australia, Brazil and France have all flagged serious concerns. That's a lot of fire for one chatbot to be standing in.

What Does Grok 4.20 Mean for AI Guardrails?

The timing of these roasts isn't random. xAI is currently rolling out the beta of Grok 4.20, and Musk has been explicit about the pitch: fewer political guardrails, better performance, less of the sanitized hedging that makes ChatGPT and Gemini feel like corporate press releases. That's the product promise. The viral roasts are, in a way, the demo.

Whether 'no guardrails' is a feature or a liability depends entirely on what Grok says next — and who it says it about.