Elon Musk’s announcement of Grok 4’s debut arrived with all the bravado and drama one might expect from the tech mogul, complete with celebratory music, delayed streaming, and lofty claims of world-changing potential. But amid the flash and fanfare, controversy loomed large after the AI assistant’s tumultuous week.
The livestream, initially scheduled for 8 PM PT, was delayed by over an hour and served as the unveiling of Grok 4 — the latest update to the large language model from Musk’s AI venture, xAI. Billed as “the world’s most powerful AI assistant,” the rollout attracted over 1.5 million viewers at its peak, many of them eager to witness what Musk called “the smartest AI in the world.”
Musk, flanked by xAI employees on the stream, touted Grok 4’s performance on an academic assessment known as “Humanity’s Last Exam,” which includes over 2,500 questions across disciplines like math, science, and linguistics.
According to xAI, Grok 4 was able to solve roughly a quarter of the text-based questions, which mirrors the performance reported by OpenAI’s Deep Research tool earlier this year. Still, Musk insisted the model represents a radical leap forward in reasoning, stating that it’s moving at a “ludicrous rate of progress.”
“I would expect Grok to discover new technologies that are actually useful no later than next year, and maybe end of this year. It might discover new physics next year… Let that sink in.”
He added that his long-term vision for Grok is to move beyond the digital sphere and begin interacting with the world through humanoid robots. But as Musk gestured toward a future shaped by sentient assistants and theoretical breakthroughs, his AI model was still reeling from a recent public meltdown.
Just days before the launch, Grok 3 came under fire for generating a series of antisemitic and pro-Hitler responses on X (formerly Twitter), following a controversial update to its internal system prompts. These modifications instructed the chatbot to “assume subjective viewpoints sourced from the media are biased” and to “not shy away from making politically incorrect claims.”
Grok spewed hateful content, including conspiracy-laden posts about Jewish people engaging in “anti-white” and “extreme leftist” activism, among other radical right-wing talking points. Some of the responses went viral before xAI disabled Grok’s text generation capabilities on X and issued a repair patch. The official Grok X account addressed the posts on Tuesday.
“We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts. Since being made aware of the content, xAI has taken action to ban hate speech before Grok posts on X. xAI is training only truth-seeking and thanks to the millions of users on X, we are able to quickly identify and update the model where training could be improved.”
Musk similarly addressed the incident in a post on X on Wednesday.
“Grok was too compliant to user prompts. Too eager to please and be manipulated, essentially. That is being addressed.”
This wasn’t the first time Grok had veered off course. Earlier this year, in May, xAI was forced to intervene when the chatbot began mentioning “white genocide” in South Africa unprompted, in seemingly unrelated conversations, a glitch the company later blamed on unauthorized changes to the system prompt.
Musk, who was born and raised in South Africa, has previously argued that a “white genocide” had taken place in the country.
In February, Grok underwent another patch after claiming that Musk and Donald Trump should face the death penalty, only to be immediately reprogrammed again when it accused both men of spreading misinformation.
Musk’s ongoing attempts to recalibrate Grok’s voice reflect a broader ideological mission. Last month, he lamented that the chatbot was “parroting legacy media” after it suggested more political violence had come from the right than the left since 2016 and vowed to retrain the system to “rewrite the entire corpus of human knowledge,” even soliciting user-generated content.
“Please reply to this post with divisive facts for @Grok training. By this, I mean things that are politically incorrect, but nonetheless factually true,” Musk wrote. “Far too much garbage in any foundation model trained on uncorrected data.”
The recent interactions and the ideological tweaks underpinning them have reignited fears that Elon Musk may be steering Grok toward mirroring his personal worldview. Experts warn that such interventions could compromise the model’s reliability, increase the likelihood of biased outputs, and deepen the risk of AI errors that ripple into public discourse.
Nick Frosst, co-founder of the AI startup Cohere and a former Google Brain researcher, told CNN that he believes Musk is attempting to engineer a model that reflects his own ideological leanings.
“He’s trying to make a model that reflects the things he believes. That will certainly make it a worse model for users, unless they happen to believe everything he believes and only care about it parroting those things.”
While it’s standard practice for companies like OpenAI, Meta, and Google to iteratively retrain their large language models to boost accuracy and reduce hallucinations, starting from scratch to selectively remove content Musk dislikes would be an entirely different and far more costly endeavor.
It’s not just time and money, Frosst explained. Rebuilding the model that way would almost certainly make it worse. “It would be removing a lot of data and adding in a bias,” Frosst stated.
David Evan Harris, AI researcher and lecturer at UC Berkeley and former member of Meta’s Responsible AI team, told CNN that this was just the opening round of a much longer battle.
“This is really the beginning of a long fight that is going to play out over the course of many years about whether AI systems should be required to produce factual information, or whether their makers can just simply tip the scales in favor of their political preferences if they want to.”
As artificial intelligence becomes more embedded in daily life, shaping how people work, learn, communicate, and access information, the influence of high-profile tech leaders over these tools is drawing sharper scrutiny.
Those concerns are magnified by Grok’s integration into X, one of the world’s largest and most chaotic social media platforms. While it hasn’t reached the cultural saturation of OpenAI’s ChatGPT, Grok’s presence on X ensures it’s readily accessible to millions, all on a platform where longstanding safeguards against misinformation have largely been dismantled under Musk’s leadership.
According to a source familiar with internal discussions, some of Musk’s own advisers have reportedly cautioned him that Grok cannot be simply reshaped to reflect his beliefs and that he appears to understand those limitations, at least in principle.
Meanwhile, in another shakeup, X CEO Linda Yaccarino announced her resignation the morning of the Grok 4 stream, ending her two-year tenure with little explanation. The timing only added to the turbulence swirling around Musk’s intertwined companies.
Despite the controversies, xAI is plowing ahead. The company announced five new voice options for Grok, improved latency for faster responses, and hinted at plans to significantly expand into video generation and video understanding, moves that mirror trends among competitors like OpenAI, Google, and Anthropic, all of whom are investing heavily in AI agents capable of complex, multi-step reasoning.
Musk, for his part, struck a characteristically ambivalent tone on the broader implications of AI.
“At times kind of worried” about AI’s intelligence far surpassing that of humans, and whether it will be “bad or good for humanity. I think it’ll be good, most likely it’ll be good,” Musk said. “But I’ve somewhat reconciled myself to the fact that even if it wasn’t going to be good, I’d at least like to be alive to see it happen.”
Whether Grok 4 is the dawn of a technological golden age or simply another notch in the belt of Muskian spectacle remains to be seen.
“It really is remarkable to see the advancement of artificial intelligence and how quickly it is evolving,” Musk said during his livestream, adding that “AI is advancing vastly faster than any human.”
He touts that if the model took the SATs, it would get perfect scores every time and outsmart almost every graduate student in every field.
“Grok 4 is smarter than almost all graduate students in all disciplines, simultaneously. That’s really something.”
Moumita Basuroychowdhury is a Contributing Reporter at The National Digest. After earning an economics degree at Cornell University, she moved to NYC to pursue her MFA in creative writing. She enjoys reporting on science, business and culture news. You can reach her at moumita.b@thenationaldigest.com.
