Karpathy Says X Has Never Been This Toxic; Musk Responds

A veteran user sounds off

AI researcher Andrej Karpathy has delivered a blunt critique of X, the platform formerly known as Twitter, and drawn an unusually candid response from its owner, Elon Musk. According to The American Bazaar, Karpathy said the site now feels more hostile than at any point across the nearly two decades he has used it.

Karpathy, who worked alongside Musk at Tesla before later stints at OpenAI and now Anthropic, did not soften his assessment. The report quotes him saying, "I've been on Twitter for almost 2 decades now so I can say with confidence that it has never been this toxic and Reddit-like." He argued that the platform's recommendation algorithm actively rewards divisive and combative behavior, and said the deteriorating climate has cut into how much he chooses to engage there.

His comment dovetails with a theme he has returned to recently: that he now spends more of his time inside collaborative work tools than scrolling a public feed. Coming from someone with a large following and deep roots on the platform, the critique carries extra weight, since it is not the complaint of a casual user but of a longtime power participant.

Musk agrees and calls for an overhaul

What made the exchange notable was Musk's reaction. Rather than defending the platform or pushing back, he appeared to concur. Per The American Bazaar, Musk replied with a direct call for change, saying, "We need a complete overhaul of the algorithm." He also restated his longstanding position favoring algorithmic transparency, contending that making X's recommendation systems more visible would strengthen accountability and help preserve public trust.

It is rare for the owner of a major social platform to publicly endorse a critic's premise that the experience has gotten worse. The concession suggests that even those steering the product share some unease about how engagement-optimized feeds shape the tone of online conversation.

The bigger picture

The back-and-forth unfolded against a backdrop of broader industry debate over AI governance and intensifying competition among leading AI firms, the report notes. Karpathy joined Anthropic in May 2026 to help advance its Claude models, a move that has kept him in the spotlight and lent his platform critique additional resonance within the AI community.

The core points of the exchange were:

• A widely respected technologist publicly questioned the health of X's discourse.
• The platform's own owner conceded that the recommendation algorithm needs rethinking.
• The moment spotlighted the persistent tension between engagement-driven feeds and constructive dialogue.

The disagreement most platforms try to avoid, between a high-profile critic and the person who runs the service, instead turned into a brief moment of agreement. That alignment underscores how even insiders are uneasy about the direction of algorithm-driven social media, where systems tuned to maximize attention can end up amplifying conflict.

What comes next

Whether Musk's stated intent to overhaul the algorithm translates into concrete changes remains to be seen, and public pledges to rework recommendation systems are easier to make than to deliver. Any genuine shift would have to balance the engagement that drives advertising revenue against the goal of a healthier conversation, a tradeoff that has bedeviled every large social network. For now, the episode stands as a candid acknowledgment, from both a prominent user and the platform's owner, that the current formula is leaving even its most invested participants frustrated.

Related on Ni4o: Karpathy's Software 2.0 Thesis Evolves Into Software 3.0