Stay up to date with notifications from The Independent

Notifications can be managed in browser preferences.

OpenAI says it is updating ChatGPT’s personality as users rage at new update

Users will now also be able to choose how much the model thinks before answering

Andrew Griffin
Wednesday 13 August 2025 16:34 BST
Comments
(AFP via Getty Images)
Leer en Español

OpenAI is working to change ChatGPT’s personality after a recent update let to outcry among users.

Last week, OpenAI unveiled GPT-5, its first major update to the system that powers ChatGPT in years. It said at the time that it would be more helpful in a range of ways, including deceiving users less and the ability to automatically pick which version of the model would answer.

But the new model was immediately met with disappointment. Analysts and critics noted that the new system did not match up to the many grand claims and hype that had preceded the launch.

Others, however, suggested something more personal had gone wrong. Across Reddit and other forums, users of the new system complained that the personality of the new GPT-5 was not as friendly or otherwise not “warm” enough.

Soon after the launch, OpenAI allowed users to switch back to the old GPT-4o model that had been available until GPT-5’s launch. OpenAI had previously suggested that the new model was so decisively improved that users would not want to switch.

Now, OpenAI chief executive Sam Altman has said that the company will improve GPT-5’s personality, apparently to address those user complaints about it feeling unfriendly.

“We are working on an update to GPT-5’s personality which should feel warmer than the current personality but not as annoying (to most users) as GPT-4o,” he wrote on Twitter. “However, one learning for us from the past few days is we really just need to get to a world with more per-user customization of model personality.”

He also said that users will now be able to choose whether to have the “fast” or “thinking” version of the model. At launch, OpenAI said that GPT-5 would be able to automatically switch between those versions depending on what would offer the best answer – and that will still be available, but users will now also be able to automatically change between them.

Join our commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies

Comments

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged inPlease refresh your browser to be logged in