Your temporary ChatGPT conversations could soon become truly private

By Nadeem Sarwar Published August 20, 2025

OpenAI has recently courted a heck ton of flak for allowing ChatGPT conversations that went seriously haywire. One person sought health advice and ended up with a form of psychosis seen in the previous century. There have been cases where the conversations even proved fatal. 

Now, the company is weighing a privacy option that could raise further alarms as it intends to lock temporary chats behind a layer of encryption. It’s a positive move for users concerned about the privacy of their chats, as it would likely prevent other parties from having a peek at the conversations, but could also prove to be a hurdle for law enforcement authorities. 

In a session with reporters, OpenAI chief Sam Altman mentioned that temporary conversations with ChatGPT could go behind encryption in the near future. When you enable the temporary chat feature, the AI chatbot has no memory of your previous conversations, and as a result, it won’t tailor the responses based on your preferences or conversation history.

It’s as if the slate has been wiped clean. These interactions do not appear in the chat history, and they’re not used for AI model training purposes either. OpenAI’s policy page, however, warns that these chats can be retained for 30 days, citing “safety” requirements. 

Now, there are a lot of reasons why one would want to have a conversation with ChatGPT in temporary mode. Think of it as using incognito mode for web browsing, but instead of surfing the internet with a browser, you are just engaged in a tete-a-tete with OpenAI’s chatbot. On paper, temporary chats sound like a safe (and more private way) of having sensitive conversations, but they’re not totally private. 

The problem with chatbot conversations and their perceived privacy begins when law enforcement comes calling. In a recent appearance on a podcast, Altman hinted at how the company is bound by the local legal framework to hand over chat records when it is duly requested. 

“So, if you go talk to ChatGPT about your most sensitive stuff and then there’s like a lawsuit or whatever, like we could be required to produce that. And I think that’s very screwed up,” the OpenAI chief was quoted as saying. That is where encryption comes into the picture. “We’re, like, very serious about it,” Altman told reporters in a recent conversation with reporters, as per Axios.

Notably, he didn’t reveal a timeline as to when encryption will be implemented for temporary chats. ChatGPT users are increasingly having intimate conversations with the chatbot, including medical and legal conversations, but as of now, none of them are protected behind a layer of confidentiality that you would expect from a doctor or lawyer. Encryption could at least partially address the privacy scares attached to temporary chats.