OpenAI has announced the rollout of memory enhancements for ChatGPT to all users, including those on the free tier. This update, disclosed on Tuesday, enables the San Francisco-based artificial intelligence company to offer personalized interactions for users based on their prior conversations with the chatbot. The free tier will experience a more streamlined version of this feature, while Codex, the firm’s AI dedicated to coding, will gain new capabilities, including full internet access during setup.
ChatGPT’s Memory Boost Feature Is Now Available to All Users
In a recent post on X (previously Twitter), OpenAI announced the expansion of its memory feature, which was initially introduced in April for ChatGPT Plus, Team, and Pro users. Now, free-tier users can also benefit from this personalized experience.
Previously, ChatGPT could remember specific user information, but modifications required users to manually adjust their settings through the Memory section. The latest upgrade allows ChatGPT to retain context from past interactions, enabling more fluid conversations without users needing to reiterate prior details.
This development means that during new sessions, users can converse normally, and ChatGPT will already have awareness of relevant context. For example, users can prompt, “Based on our conversation about the quarterly performance report, what should our goals be for the next quarter?” and the model will respond appropriately.
While OpenAI has confirmed that free users will experience a toned-down version of the memory feature compared to paid subscribers, details on the specific differences have not been disclosed.
In addition to the memory updates, OpenAI is enhancing its coding agent, Codex. According to a separate post by OpenAI CEO Sam Altman, Codex will now have the ability to access the internet, an option that can be enabled during the setup process. However, this feature will be off by default.
OpenAI cautions that granting internet access to Codex presents potential security risks, as stated in their support page. Risks mentioned include prompt injection, code exfiltration, and the possibility of malware exposure. These dangers may arise as the AI agent navigates the web and processes unverified content.
Moreover, Altman noted that Codex is now available to ChatGPT Plus users, expanding access from its prior availability solely to Pro subscribers.
