OpenAI has initiated the rollout of enhancements to ChatGPT’s memory feature for users on the free tier. The San Francisco-based artificial intelligence company announced this update on Tuesday, revealing that now all ChatGPT users will receive personalized responses derived from their prior interactions with the chatbot. However, free users will only experience a simplified version of this memory feature. Additionally, the firm is allowing Codex, its coding-oriented AI agent, full internet access that can be enabled during the setup process.
ChatGPT’s Memory Boost Feature Now Accessible to All Users
In a post on X (formerly Twitter), OpenAI’s official account announced the introduction of this new feature. Initially, the memory boost was available only to subscribers of ChatGPT Plus, Team, and Pro plans since its launch in April, but it is now being extended to users on the free tier.
Previously, ChatGPT could save and recall specific user information, which could be modified or deleted via the Memory section in Settings. The latest upgrade allows the AI chatbot to acquire context and information from earlier dialogues with users.
This enhancement means that users no longer need to provide extensive details in prompts at the beginning of new sessions. They can engage in regular conversations, and ChatGPT will already be informed about the relevant context. Users can also reference previous discussions using prompts like, “Based on our conversation about the quarterly performance report, tell me what the goals should be for next quarter,” allowing the chatbot to comprehend the context effortlessly.
OpenAI noted that free users will receive a lighter version of this feature compared to the full experience offered to paid subscribers. However, the company has not specified what those differences entail.
In a separate announcement, OpenAI is also enhancing its coding AI, Codex. In a post on X, CEO Sam Altman stated that Codex will now have unrestrained access to the Internet, which users can activate during the setup stage. By default, this feature will remain disabled.
On its support page, OpenAI has emphasized the potential security risks associated with enabling internet connectivity for Codex. Such risks may include prompt injection, code exfiltration, and the possibility of encountering malware, as the AI agent interacts with unverified content across the web.
Furthermore, Altman indicated that Codex is now also accessible to users of the ChatGPT Plus tier, expanding availability beyond the previous Pro subscribers only.
