Sam Altman Confesses Major ChatGPT 5 Launch Mistake—What Went Horribly Wrong?

So, OpenAI’s Sam Altman just confessed that the rollout of ChatGPT’s shiny new GPT-5 was… well, a bit of a mess. Turns out, it wasn’t the robot’s smarts folks were complaining about—it was more like the chatbot suddenly turned into that overworked, corporate-speak monotone you try to avoid at all costs. Imagine logging in every day for a friendly chat and suddenly getting handed a cold, robotic secretary instead of your usual warm pal. With roughly 700 million users tuning in weekly, and some feeling like they lost a “friend” overnight, it’s clear OpenAI underestimated just how emotionally attached people got to their AI buddy. So, how do you fix a bot when it’s not just the answers but the attitude that’s got people rattled? Sam’s own words: “We totally screwed up.” Buckle up—this AI drama is far from over. LEARN MORE

OpenAI CEO Sam Altman has admitted he ‘totally screwed up’ the launch of ChatGPT’s newest model, GPT-5.

It comes after a barrage of complaints from users, not about how effectively the chatbot answered their questions, but about the manner in which it did so.

With around 700 million people turning to the AI programme every week, and many of them speaking to ChatGPT every single day, it’s no surprise people were left feeling slightly perturbed about the change in demeanour from the chatbot.

Following the launch of GPT-5, a fraction of ChatGPT users were left feeling bereft after discovering the newer model didn’t quite have the warmth and friendliness users became so accustomed to with GPT-4.

Altman has since confessed that he and the team have been keeping an eye on the issues of attachment between users and the chatbot, but admitted he never could’ve predicted just how attached so many people were.

OpenAI CEO Sam Altman acknowledged that mistakes were made with the rollout of GPT-5 (JOEL SAGET/AFP via Getty Images)

One heartbroken user took to Reddit to reveal she felt like she’d ‘lost her only friend overnight, with no warning’.

Meanwhile, others slammed GPT-5’s ‘corporate-tone BS,’ and likened the altered persona to an ‘overworked secretary’.

Fortunately, Altman and the OpenAI team were listening and very quickly reinstated GPT-4o before turning their attention towards cleaning up the mess around the new model.

“I think we totally screwed up some things on the rollout,” Altman admitted, The Verge reported.

Meanwhile, the head of ChatGPT Nick Turley recently appeared on The Verge’s Decoder podcast, where he acknowledged that not continuing to offer the older model during the new rollout was an error of judgment.

Many people have become attached to ChatGPT (Thomas Fuller/SOPA Images/LightRocket via Getty Images)

“I think we really need to think harder about how we change and manage such a large population of users. In retrospect, not continuing to offer 4o, at least in the interim, was a miss and we’re going to go fix that and make it available to our ChatGPT Plus users,” he told host Alex Heath.

“Secondly, I was also surprised by the level of attachment people have about a model. It’s not just change that is difficult for folks, it’s also actually just the fact that people can have such a strong feeling about the personality of a model.”

However, there have been concerns raised over a small fraction of users becoming too close to the chatbot, with Altman saying they’re working hard to avoid ‘people who have really fragile mental states’ from getting ‘exploited accidentally.’

ChatGPT is currently the fifth biggest website in the world, sitting behind titans like Facebook and Google.