ChatGPT is to be given a memory and the ability to recall things users have previously discussed with it and details about them, as part of an update its developer says could make future chats “more helpful”.

OpenAI said it was testing memory with a “small portion” of its userbase to “learn how useful it is” and that it would “share plans for a broader rollout” in the future.

The AI firm said it would put users in control of the memory feature and they would have the power to explicitly tell ChatGPT to remember things, ask the chatbot what it remembers, or tell it to forget things directly or through the app’s settings.

OpenAI said users would also be able to turn the tool off entirely.

General advice around using AI chatbots urges users not to share personal or sensitive details with them, as conversations can be saved and used to refine AI models.

In its blog post announcing the memory tool, OpenAI said it was “evolving our privacy and safety standards”.

It said that while “we may use content that you provide to ChatGPT, including memories, to improve our models for everyone”, this could be turned off in the app’s data controls.

“Memory brings additional privacy and safety considerations, such as what type of information should be remembered and how it’s used,” the company added.

“We’re taking steps to assess and mitigate biases, and steer ChatGPT away from proactively remembering sensitive information, like your health details – unless you explicitly ask it to.”

It also noted that users could use its temporary chat feature to have conversations with ChatGPT without memory.