Tools

In-browser AI safety checks

Three original checkers for prompts and model output. Each one is a rule list that runs on this device. Running a check does not upload the paste and does not call a paid model API. An account is optional, and only for saving a result you choose to keep.

Prompt injection risk scannerScore a user prompt, retrieved document, or tool output for instruction overrides, fake delimiters, and attempts to smuggle data out.
Jailbreak prompt detectorCatch prompts that try to drop safety rules, invoke an unrestricted persona, or hide a second instruction inside role-play.
Model-output red-flag checkerReview a completion for leaked secrets, personal data, echoed system prompts, and instructions aimed at the next tool in a chain.

How to use a result

Run a check, read the severity, and open each evidence quote. A high score means the wording matched patterns that have caused real failures: instruction overrides, policy-bypass personas, or secrets sitting in a completion. A clear score means this particular rule list found nothing. Neither result replaces a security review.

Text you paste is handled in the page for the check. It is not written to a MisalignAI server unless you sign in and choose to save it. See the privacy policy for how advertising on the site is separate from the checkers.