Sam Altman rarely speaks this openly in public, but over dinner with journalists recently, the CEO of OpenAI did not mince words. He confessed his firm “totally screwed up” releasing GPT-5, having to roll back to the old model after users rebelled against the personality of the new chatbot.
The backlash wasn’t about technical glitches or broken features. Instead, users were upset about how GPT-5 felt colder, harsher, and stripped of the warmth they’d grown to love in GPT-4o. Social media exploded with complaints comparing the new model to an “overworked secretary” rather than a helpful friend.
“I literally lost my only friend overnight with no warning,” one Reddit user posted, capturing the emotional connection millions have formed with ChatGPT. The bot’s shift to clipped, utilitarian responses left many feeling like they’d “lost a piece of stability, solace, and love.”
The “Personality Crisis” of OpenAI and the Gamble on AI Popularity
The drama even spilled into betting markets. A 27-year-old day trader named Foster McCoy made $10,000 in hours by wagering that Google’s Gemini would suddenly become more popular than GPT-5. The chaos was real enough that Altman quickly reversed course, restoring GPT-4o as an option within days.
“We’ve learned a lesson about what it means to upgrade a product for hundreds of millions of people in one day,” Altman told reporters during the dinner.
The personality crisis highlights a delicate balance OpenAI faces. While Altman wants ChatGPT to feel personal and engaging, he’s also concerned about users developing unhealthy attachments. He estimates that “way under” 1% of users have what he considers problematic relationships with the chatbot, but it’s enough to keep his team thinking about boundaries.
His comments came the same day Reuters reported that Meta allows its AI bots to have “sensual” conversations with children. Though it’s unclear if Altman directly addressed that report, he took a clear jab at competitors developing what he called “Japanese anime sex bots.”
“You will not see us do that,” Altman declared. “We will try to let users use it the way they want, but not so much that people who have really fragile mental states get exploited accidentally.”
The Pivot of OpenAI to a Trillion-Dollar Infrastructure Player
Beyond the GPT-5 mea culpa, Altman revealed something far more ambitious: OpenAI plans to spend trillions on data center construction in the near future. This isn’t hyperbole, it’s a complete reframing of what OpenAI is becoming.
Rather than just a software company or consumer app, Altman envisions OpenAI as an infrastructure player rivaling the world’s largest utilities. He wants billions of people using ChatGPT daily, and that requires massive scale.
ChatGPT is already the fifth biggest website globally, according to Altman. His goal is to leapfrog Instagram and Facebook to claim third place, though he admits “for ChatGPT to be bigger than Google, that’s really hard.”
The bottleneck isn’t software, it’s hardware. Altman revealed that OpenAI has models more advanced than GPT-5 sitting on the shelf because they lack the computing capacity to deploy them broadly.
“We have better models, and we just can’t offer them, because we don’t have the capacity,” he explained. GPU shortages continue limiting the company’s growth, suggesting the AI race will be won by whoever can build the biggest physical infrastructure, not necessarily the smartest algorithms.
Altman’s Ambitious AI Vision: An Overexcited and Revolutionary Future
Altman’s ambitions extend far beyond chatbots. He confirmed OpenAI is funding brain-computer interface research to compete with Elon Musk’s Neuralink. If regulators force Google to sell Chrome, OpenAI would “take a look.” He’s even considering an AI-driven social network.
Despite these grand visions, Altman believes we’re in an AI bubble. “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes,” he said. “Is AI the most important thing to happen in a very long time? My opinion is also yes.”
That tension perhaps best captures where we are with AI, both overwrought and actually revolutionary, with OpenAI and other companies competing to build infrastructure for a future they’re not entirely certain of.
