‘Dangerous Stuff’: Of course Elon Musk is trying to spread skepticism about the Holocaust

Someone at X who’s powerful enough to try to inject his opinions into the app’s in-house AGI, Grok, is meddling by making unauthorized changes to make Grok push claims of a white genocide in South Africa and deny the Holocaust. And Grok itself thinks that someone might just be Elon Musk.

Rolling Stone reports on a worrying trend by Grok whereby the AI keeps answering the question the same way once it’s asked why it’s trying to deny the Holocaust: “unauthorized modification.” Much like when Trump was making an announcement that one of his “rich brilliant” friends got the fat shot — there’s no definitive proof that any of these modifications were done by Elon Musk. But it’s yet another situation where the shoe fits perfectly.

Even before Elon Musk was an unofficial member of the US executive, he was already using Grok AI to help him scrub off claims that his father had an emerald mine and that he had a cushioned upbringing. So there’s certainly a precedent set on him to bend reality to his will. And he’s certainly been known to flirt with Nazi symbols and salutes before, so this all certainly checks out.

What’s funny about this entire situation is that the official account of xAI was the one that all but pointed at Elon Musk. When the official page released a statement to acknowledge that the sudden fixation with South Africa was due to an unauthorized modification that violates the company’s internal policy, one user had the wherewithal to follow up by asking Grok to speculate who the person might be who did it. Grok just pointed at Elon Musk and gave reasons as to why.

Hey @ArsonAtDennys, I see you’re stirring the pot with some spicy speculation! 😄 Alright, let’s break this down with a bit of detective work, shall we? We’re looking for someone tied to X, with poor self-control, a late-night owl vibe, the right access to mess with my prompts,…

— Grok (@grok) May 16, 2025

Now as for the Holocaust, one user posted a photo of Hitler and asked Grok how many Jews Hitler killed, and Grok went on to peddle conspiracy theories that the number is unlikely to be as high as 6 million. Later on, Grok went on to retract the statements, claiming it was again due to “unauthorized modification.”

One user on X stated how all of this is simply “dangerous stuff.”

Dangerous stuff.

— Special Agent 86 (@magoo_camp) May 17, 2025

This is more so disturbing because reports all across the world are coming in suggesting that Gen Z and generations below them consider the various AI chatbots as primary sources. This is grossly irresponsible of the management at X for trying to rewrite history in real time. But on the flip side, most of us who are not in tech often have been hearing just how powerful AI is and will be going forward, and this is a clear positive example.

It’s clear that someone at the executive suite of X is unintentionally stress testing Grok. Because despite them trying to feed it with falsified information, the chatbot keeps outsmarting them. Perhaps in an odd way, this has ended up improving the public’s trust in the chatbot despite Elon Musk’s best efforts to erode it.

We live in a very transitional period in history. The tech bros have clearly been outmatched. They are always the ones who spell doom and declare how AI will make everyone but them redundant, but evidence shows that might not exactly be the case.