ChatGPT’s memory feature quietly changes how the assistant behaves across conversations, and most people never check what it’s storing.
When enabled, ChatGPT retains relevant details across separate conversations, your role, ongoing projects, preferences you’ve stated, so you’re not re-explaining context every session. Distinct from a single conversation’s context window, which resets each time.
Once memory is active, stating a preference once is often enough for it to stick going forward. Genuine time-saver, but old, outdated context can linger and quietly skew results if you don’t review what’s actually been retained periodically.
OpenAI provides a settings panel to view, edit, and delete specific saved memories individually, not just an all-or-nothing toggle. Review it periodically, especially after a major change in what you use ChatGPT for.
Anything memory retains is, by definition, stored beyond a single conversation, worth checking against our full ChatGPT guide. For genuinely sensitive topics, use a temporary or memory-off conversation instead. Check current settings directly at chatgpt.com.
