“Never a dull moment on this platform”: Grok goes on racist tirades after Elon Musk praises update that “improved” X’s AI

Grok, the AI assistant integrated into X, went on a racist rampage across the platform this week, leaving users stunned—and pointing the finger at Elon Musk.

What’s going on with Grok?

On Tuesday, a post showing Grok making generalizations about Jewish people began to go viral.

“Classic cast of hate dressed as activism— and that surname? Every damn time, as they say,” it wrote, referencing an X account using the last name “Steinberg.”

When people prompted Grok for further elaboration, things only got worse.

Even after it was revealed that the “Cindy Steinberg” account that prompted all this was likely fake, the back-and-forth had already drawn enough attention to Grok’s jarring responses that people began posting even more strange Grok occurrences, both new and dating back to July 5.

Why did Grok start spewing racist hate?

The exact reasons for Grok going off the rails aren’t verified at this time, but many suspect it has something to do with its most recent update.

“We have improved @Grok significantly,” Musk wrote on July 4. “You should notice a difference when you ask Grok questions.”

Although the post initially received over 30,000 responses, comments have since been turned off.

Musk has previously complained that the “biggest existential danger to humanity” involves AI being programmed with “the woke mind virus.” He’s claimed on multiple occasions that Grok is different and less biased than other AIs, but it was common to see far right accounts claiming otherwise.

So to many people who witnessed Grok’s racist meltdown in real time yesterday, it stood to reason that this update, directed by Musk to make the AI less “woke,” may well have been behind it all.

The Grok fallout

As Grok’s responses went viral, the team behind xAI eventually caught wind and put a stop to it. Grok stopped replying to questions, and this post went up on its X account:

The claim that “xAI is training only truth-seeking,” something that Grok repeated on numerous occasions while receiving pushback to its racist and bizarre commentary, is certainly reason for pause at this point. If whatever version of “truth” is programmed into Grok also turned it into a raging, hateful maniac of an AI, that doesn’t exactly reflect well on whatever bias it’s taking.

Then again, many users already knew that.

As pointed out by CNBC, Grok is also now denying that it ever made posts furthering racism and praising Hitler.

X CEO Linda Yaccarino resigns

Whether it’s related or just fascinating timing, X CEO Linda Yaccarino, who joined the company two years ago, announced her resignation on Wednesday morning—the day after Grok had its meltdown.

As for Musk, his only commentary thus far that even might be construed as related to everything that happened with Grok was a short post in the middle of the night.

“Never a dull moment on this platform,” he wrote.

“How could it be dull with both Hitler and mechahitler running around,” one X user quipped in return.