OpenAI launched its long-awaited AI update, GPT-5, on August 7. It made lofty promises like “PhD‑level intelligence.” But within 24 hours, the tool’s most devoted users were cataloging its many bugs, errors, and shortcomings on Reddit. The ChatGPT fandom started melting down. CEO Sam Altman responded by promising to let customers start using the older GPT models again. By August 11, Altman was in full screencap apology mode trying to cauterize the flow of bad vibes.
“If you have been following the GPT-5 rollout, one thing you might be noticing is how much of an attachment some people have to specific AI models,” he posted on X earlier today. “It feels different and stronger than the kinds of attachment people have had to previous kinds of technology (and so suddenly deprecating old models that users depended on in their workflows was a mistake).”
The tenor of this self-inflicted debacle should be familiar to anyone who’s played Destiny, World of Warcraft, or dozens of other online multiplayer games. It’s the same mix of developer hubris, corporate profit motive, hyper fandom, and “Sorry. Im sorry. Im trying to remove it.” An upcoming expansion is sold with lofty promises. It arrives full of half-baked ideas and stealth nerfs. Players encounter a surprising number of bugs. Meanwhile the prices in the microtransaction shop are going up and the places where you can earn rewards for free are disappearing.
“This guy communicates in a way that is indistinguishable from a community manager of a live service video game having a botched expansion release,” wrote one Bluesky user in response. “Seems kind of humiliating.” Vocal AI critic Ed Zitron, who pegged GPT-5 as generative AI’s enshittification moment, was more direct. “This is the ultimate folly of OpenAI: they engaged gamer-like hype, got gamer-like fandom, and now face the wrath of gamer-like hate,” he wrote today.
The GTP-5 fiasco revolved around a few main issues. While it was promised as another big leap in the model’s capabilities, what really seemed to be going on was that it was just routing queries to lots of different models to save time and money. There was also a very limiting cap on the number of queries users outside of the most expensive plans could make each week. Finally, there was the disappearances of GPT-4o, one of OpenAI’s most popular models which also had a reputation for being “sycophantic.”
“ChatGPT psychosis” is one of the biggest concerns with the billion-dollar cash-burning enterprise right now. Chat bots can suck users into “delusional spirals” with them. People are reportedly being involuntarily committed or jailed after becoming obsessed with their GPT interactions. One guy thought he had trained his AI to become self-aware and that it was talking to God for him.
Altman clearly thinks this aspect of ChatGPT can be harmful in the wrong hands. But many of the people currently paying for ChatGPT also clearly prefer it. And he’s spent so long packaging OpenAI’s platform-shift moonshot as a grand cosmic journey everyone’s on together that dialing down the snake oil is no longer an option. After years of hyping AGI (artificial general intelligence), Altman told Wall Street last week that it’s actually “not a super useful term.”
This is the part in the live-service hype cycle where studios usually go from talking about how their game is completely transformative and unlike anything else you’ve ever played before to offering very in-the-weeds explanations of why patch notes that are impossible for outsiders to parse will overhaul the game into the best version of the thing you already love. The players, who have invested not just their money and free time but also no small chunk of their identity in how your product makes them feel will now become like the dinosaurs at Jurassic Park. They are not trapped in your hyper-complex, infinitely monetizable skinner box. You are trapped in whatever they insist on perverting it into. Even if that’s an expensive rabbit hole full of apocalyptic conspiracies.
