Everyone’s job is safety: Elon Musk fires back at xAI exodus concerns

Elon Musk is being criticized for ignoring safety measures after former employees reported that he dismantled the internal safety department at his xAI startup, responsible for the Grok chatbot. 

Only 12 of xAI’s original cofounders remain employed at the company after several of them left for various reasons, including to start their own companies and complaints of creative stagnation.

Is xAI sacrificing safety to compete with OpenAI?

One source who spoke to The Verge claimed that xAI’s safety team has been effectively dissolved, saying “Safety is a dead org at xAI.”

According to reports, there’s been a push for “unfiltered” content, leading to a focus on NSFW (Not Safe For Work) capabilities for the Grok AI.

Former staffers allege that Musk views safety measures as a form of “censorship.” They claim that engineers are encouraged to “push to production” immediately, sometimes bypassing traditional testing phases.

This culture has reportedly led to internal friction between leadership, causing them to clash often over product priorities in large group chats on the X platform.

Musk argued on X that “everyone’s job is safety.”

Using Tesla and SpaceX as examples, he said that neither company has a massive, independent safety department, yet they produce the safest cars and rockets in the world.

To Musk, separate safety departments are often “fake” and exist only to “assuage the concerns of outsiders” without having any real power to improve the product.

In Musk’s ongoing legal battle with OpenAI and its CEO, Sam Altman, he has frequently criticized OpenAI for becoming a closed-source for-profit company that prioritizes profit over safety, but he is being accused of doing the same by removing the internal checks and balances that prevent AI from generating harmful or biased content.

Frequent hiring at xAI

Following the recent announcement of a merger between xAI and SpaceX that led to an internal valuation of approximately $1.25 trillion, the company has seen a wave of high-profile departures causing people to question the company’s internal culture, its technical direction, and its approach to AI safety. Former employees describe the company as a “dead” safety organization.

Two of the company’s most prominent co-founders, Yuhuai (Tony) Wu and Jimmy Ba, recently announced they were leaving. Wu stated it was “time for his next chapter,” while Ba noted he needed to “recalibrate his gradient on the big picture.” Only half of the original 12 co-founders who launched xAI remain at the company. Several other engineers and staffers have also resigned, with many stating they intend to start their own AI firms. One group of former employees has already launched “Nuraline,” a startup focused on AI infrastructure.

Some departing employees like Vahid Kazemi, suggested that the industry has become stagnant, writing on X that “all AI labs are building the exact same thing.” Others suggested that xAI is stuck in a “catch-up phase,” merely trying to replicate what OpenAI and Anthropic achieved a year ago rather than innovating.

In an internal all-hands meeting, Musk explained that xAI would now be divided into four primary sectors namely Grok Main and Voice, Coding, and Data Macrohard.

xAI’s Colossus supercluster in Memphis, Tennessee currently houses 100,000 Nvidia H100 GPUs, and is currently being expanded to 200,000 GPUs. This hardware is essential for training “Grok 3,” which Musk believes will surpass all other AI models currently on the market.

Get 8% CASHBACK when you spend crypto with COCA Visa card. Order your FREE card.