Write permission is access that lets a system create, update, or delete records, not just view them. In security terms, it is a higher-risk privilege than read access because it can change the source of truth. A user account, API token, service, or AI agent with write permission can alter databases, configuration, logs, or personal data.
This matters because attackers often try to turn read-only access into write access, or steal credentials that can modify records. Once write permission is abused, the impact can include fraud, account takeover, data corruption, or unauthorized changes that are hard to detect. Defenders limit this risk with least privilege, separate read and write roles, approval workflows for sensitive changes, strong logging, and rollback or versioning so bad writes can be traced and reversed. For AI agents, write permission should be tightly scoped and closely monitored.