What it does
Separate message and document checks return Jev probabilities, so the surrounding application still controls the final policy.
Integrations
Chia AI guard
Chia's AI guard scores prompt injection and inappropriate content before another model reads the text.
Bookmark: Chia AI guard
Keep this in your collection.
Leave a noteWhat would you try with this?
: Chia AI guard
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
You can screen incoming text or uploaded documents before another tool reads them. Start by writing out the exact questions you want answered, such as whether a message tries to override rules. Decide what your app should do if a check fails.
A developer can adapt this approach by connecting your software to TypeSafe using an account access key. Because the text gets sent to an outside service, keep sensitive information in mind. Automated safety scores can still make mistakes, so they do not guarantee complete safety.