Elon Musk put the number of times he’d experienced “AI or SI psychosis” this year at “probably like half a dozen times.” His remarks weren’t a report of a diagnosis or a description of delusions. Shortly beforehand, he’d expressed admiration for AI-generated music and surprise that a model seemed capable of good taste.
Musk made the comments during Saturday’s X Takeover event, according to Business Insider’s reporting. “I’ve got a little bit of cyber psychosis,” he said, following that with: “Maybe a lot actually.”
“At various points this year,” he continued, “I’ve had AI or SI psychosis probably like half a dozen times.” That tentative count comes from Musk himself; it doesn’t demonstrate six clinically established episodes. Nothing in the reporting establishes a diagnosis.
What he’d just said about music offers more context for the remark. “Calling it slop-core is a misnomer, frankly it’s awesome-core,” Musk said. “AI music is really good.”
“It just shows taste,” he asserted. “And youre like, ‘wow, AI can do good taste? That’s pretty wild.’”
At issue is the distance between the loaded mental-health term and the example he supplied. Admiring music made by AI isn’t the same as interacting with a chatbot in ways that contribute to losing contact with reality.
For some experts, “AI psychosis” describes situations in which chatbots fuel or worsen a user’s delusions, at times leading to a break with reality. The most extreme cases can involve severe mental-health spirals, disrupting both personal and professional life and sometimes resulting in self-harm or suicide. These potential consequences differ sharply from Musk’s description of surprise at AI’s apparent taste in music.
Within tech culture, the expression has another, less literal use: mocking people who become absorbed in AI and believe extravagant promises about what it can do, especially CEOs. Some respond by adopting the label with ironic pride while brushing off skeptics as Luddites. In Futurism’s reading, Musk appears to be making that sort of proud declaration, not disclosing a medical condition.
The abbreviation adds a separate complication. By “SI,” Musk meant “superintelligence.” Futurism reports that President Donald Trump had just passed an executive order replacing the term artificial intelligence with that wording. The publication sarcastically suggests that treating this name change seriously despite the technology’s obvious flaws might itself indicate disordered thinking. That argument is criticism from Futurism, not a clinical judgment.
That tone extends to Futurism’s wider assessment of Musk. The publication connects his comments to his self-glorifying AI-generated image posts and his conviction that Grok, his “maximum truth-seeking AI,” will reveal the secrets of the universe. It also points to his acceptance of conspiracy theories it characterizes as disconnected from reality, including a fringe theory concerning James Talarico.
Others close to Musk have spoken positively about the label, too. Claire “Grimes” Boucher, with whom he shares children, described supposed benefits in a tweet in December: “The thing about AI psychosis is that it’s more fun than not having AI psychosis,” adding: “I’ve had it (might still have it).” To Futurism, her comments appeared sincere, rather than an exercise in irony.
As an example of a tech figure facing allegations of actual, non-“fun” AI psychosis, Futurism points to OpenAI investor Geoff Lewis, describing the case as infamous. Its account provides neither the details of those allegations nor their outcome.
Futurism’s closing judgment is sarcastic: after hearing Musk praise AI music, it would almost worry more if he didn’t claim AI psychosis. Beneath the joke lies a more serious concern. Applying one label both to wonder at AI-generated music and to experiences of delusions and serious harm risks obscuring the differences between them.
