Cross-session memory is information retained from one chat or session and reused in a later one. In AI systems, this can include preferences, project details, identifiers, or other context that the service stores beyond the current conversation. It improves continuity and reduces repeated prompting, but it also means the account now holds more persistent state than a normal stateless chat.
In cyber security, persistence changes the risk profile. If an attacker gains account access, they may learn sensitive retained context or use it to manipulate future responses. If stale or unwanted memory is not reviewed, it can leak into new chats and cause data exposure, confusion, or policy violations. Defenders treat cross-session memory like any other stored data: minimize what is saved, separate sensitive work from casual use, and provide controls to inspect, delete, or disable remembered context. Good memory hygiene is part of account security.