Safe use of AI at work: Strong, Solid, or Needs work

The three things it checks

The rating comes from three questions, asked after your live task:

Client data. How you treat private or client information when you use AI tools.
Checking output. Whether you review AI output before it goes to a client, a teammate, or the public.
What not to hand off. What you refuse to leave to AI alone (judgment calls, sensitive decisions, work that still needs a human).

Your answers get a rating of Strong, Solid, or Needs work. You see it on your results page. It is separate from L1 to L4.

Why the rating does not change your level

Your level (L1 Novice to L4 Master) measures what you can produce with AI on a live task, plus your interview and, for L3 and L4, proof of your own work. Full rules: How scoring works.

Safe use is a different signal. Someone can produce strong work and still need better habits with data or review. Someone can be careful and still be early on AI skill. AgentProof keeps those two scores apart so neither hides the other.

What Strong, Solid, and Needs work mean

Strong: Clear, specific answers on all three questions. You treat client data carefully, you check AI output before it goes out, and you know what stays with a human.
Solid: Good habits overall, with room to get sharper on one of the three areas.
Needs work: Gaps on one or more of the three. The rating flags that for you. It does not block your certificate or lower your level.

The rating is for you and for anyone you share results with in a team setting. Individual answers stay private. More on privacy and what verify shows: how verify works.

How this fits the assessment

Four steps, about 15 minutes, all in chat:

Chat about how you work.
Do one live task (about three minutes), then the three safe-use questions.
Prove it's you (live selfie and confirmed email).
Share your proof (verify link, LinkedIn, badge).

You use any AI tools you already use. AgentProof is vendor neutral. Overview: What is AI readiness? Levels: levels glossary.

For teams and managers

Teams that run AgentProof get a private report of totals (opens at three people). The report covers group-level signals, not individual chat answers. If you care how your group handles data and review, the safe-use rating is part of that picture alongside levels. See how AgentProof works for teams.

Hiring managers who open a verify link see the certificate result. Use safe-use as one more signal alongside your usual process, not as a standalone pass/fail. More: AI readiness for hiring managers.

Quick answers

What is the safe-use rating?
A rating of Strong, Solid, or Needs work from three questions on client data, checking AI output, and what not to hand to AI. It does not change your level.

Does Needs work block my certificate?
No. You still get your level and your free certificate. The rating is a separate signal.

Is safe use part of AI readiness?
Yes. AgentProof treats skill and safe habits as both part of readiness. You finish with a level and a safe-use rating.

Do managers see my safe-use answers?
No. Your individual answers stay private. Teams see totals in a private report, not your chat answers.

When do the three questions happen?
Right after the live task, before the selfie and email step.

Does the rating expire?
No. Recheck any time if you want a fresh assessment. Your verify page shows the date you earned the certificate and the date you last checked, if you rechecked.

More in the AgentProof FAQ. Related: What is an AI skills certificate?

Get assessed

AgentProof is live and invite only. We review requests by hand each week and email you from hello@joinagentproof.com when a spot opens.

Request access. It's free for individuals.

Bringing a team? Start free for 30 days with a small group. Or read how AgentProof works for teams.