OpenAI has announced the gradual rollout of a new memory feature in ChatGPT, designed to make the AI assistant more personalized and context-aware.
The update allows ChatGPT to reference information from users’ past interactions, enabling more fluid and relevant conversations.
The feature, listed in settings as “reference saved memories,” enhances ChatGPT’s responses across text, voice, and image interactions. According to OpenAI, it aims to eliminate the need for users to repeat information, making the chatbot more conversational and efficient.
Initially, the rollout will be limited to ChatGPT Pro and Plus subscribers, with users in the U.K., EU, Iceland, Liechtenstein, Norway, and Switzerland excluded for now. OpenAI cited the need for additional external reviews to comply with local privacy regulations in those regions, but emphasized that the feature will be available there eventually. For free-tier users, there’s no launch date yet.
While the update promises a more tailored user experience, OpenAI has acknowledged privacy concerns. Users can opt out of the feature at any time via ChatGPT’s settings and manage or delete specific saved memories. They can also ask the chatbot what it remembers or use a “Temporary Chat” mode to keep conversations off the record.
OpenAI had previously introduced memory controls in 2023, but those required users to explicitly request memory changes. The new system is more automatic, offering a seamless experience — though it remains fully user-controllable.
The move follows similar developments from competitors. In February, Google added memory capabilities to its own AI assistant, Gemini, as tech firms continue to compete in the race to build smarter, more personalized AI.
