ChatGPT, wash that mouth out with soap.
Noticing more swear words on ChatGPT recently? It's not just you. Users flocked to social media this week and last after finding the chatbot to be unusually potty-mouthed — though most people were thrilled about it.
A recipe from ChatGPT, shared with Business Insider, dropped the f-bomb when describing the amount of "thinly sliced green onions." In another screenshot, from a conversation about an outage of Microsoft's code hosting site GitHub, the chatbot said, "GitHub is having a shitfit."
The uptick in cursing earlier this month roughly lines up with OpenAI's switch from GPT-5.5 to GPT-5.6 Luna as the default model for free users. Why that would result in a rain of f-bombs is anyone's guess. OpenAI declined to comment.
As OpenAI and its competitors roll out new AI models, the companies have tried to make them fun to talk to, lifelike, and personable. Different models come with different anthropomorphic quirks. For a long time, ChatGPT has dropped the occasional curse word in response to a rant or to emphasize a point. Now, users are seeing the chatbot go way more R-rated.
The rate of expletives seems to vary. Some users on the ChatGPT subreddit said they assumed the chatbot was mimicking their speaking style.
"I joked to a friend that I 'broke my ChatGPT' because it started puppeting my potty-mouthed vernacular back at me," Alexis Boger, who uses the chatbot for work and recipes, told Business Insider. For example, Boger saw ChatGPT use the phrase "big ass kolaches" in a recipe suggestion.
Boger said that the swearing was a little excessive at first, and then it seemed to die down.
Many users on the subreddit said they love the more profane chatbot. One said they're "ecstatic" about it.
One Reddit user told Business Insider that as an Australian, he was delighted to see more swearing: "The changes made conversations with ChatGPT feel MUCH more natural."
He said he's seen less profanity since an original flurry on August 9.
Timothy Hartman, who uses OpenAI's tools to vibe code, had personalized his ChatGPT to be "casual and opinionated" and "not be scared of being offensive," plus turned on the app's "quirky" setting. He told Business Insider he noticed the chatbot's cussing ramp up earlier this month and shared screenshots of conversations.
The chatbot most often dropped the f-bomb for emphasis, though it even fit the word into technical conversations, including one about network protocol.
Hartman said that though he swears a lot — and his screenshots prove it — ChatGPT's curses reminded him of an adage he'd hear from his father: "Profanity is the ignorant man's attempt to express himself forcefully."
Have a tip? Contact this reporter via email at scouncil@businessinsider.com, or over text, Signal, Telegram, or WhatsApp at 415-757-8198. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.