Use AI safely by de-identifying data and minimizing risks
To minimize risk, adopt privacy-preserving workflows. First, understand that consumer models often log your prompts by default, potentially using them for future training. Education-specific models, under district contract, typically disable this logging.
If you must use a personal tool, practice data minimization—only enter the information necessary to get the job done. Use aliases or placeholders instead of real student names. Before uploading a file (like a rubric or sample essay), strip any metadata (like student names embedded in the file details) that could identify the student. Always verify the settings in your account; disable history or training where possible to maintain maximum privacy
