WHAT YOU'LL ACHIEVE
Decide whether AI may be useful for a task, and what to check before you try it, without treating a result as a safety or compliance verdict.
A PRACTICAL DECISION AID
Start with the task, then the stakes.
Answer the questions to get guidance. It helps you decide; it isn't a verdict.
Answer each question to see a plain-English starting point. Your task description is optional and is only used in this page; nothing is sent or saved.
Last reviewed
YOUR STARTING POINT
A judgement aid, not a verdict.
Complete the assessment to see a result. No answers are preselected, so this neutral state does not imply that the task is safe or suitable for AI.
This is educational guidance, not compliance certification. Organisational policy and product terms govern.
Rules: high impact if wrong, consequential decisions about people, or inability to verify always means “Do not rely on general-purpose AI alone.” Personal or confidential information, moderate impact or partial verification means “Potentially useful with safeguards.” The first class requires public or non-sensitive information, low impact if wrong, independent verification and no consequential decisions. Task type and repetition are included in the rationale but do not lower risk.
KEEP GOING
A useful next check.
Before opening a tool, look at how to share information carefully and compare tools against the actual job, access and rules that apply to you.
Leila Fielding