I can see this being used to detect accidentally leaked sensitive content (for example scanning repos, outgoing emails, shared folders). However using this to

I can see this being used to detect accidentally leaked sensitive content (for example scanning repos, outgoing emails, shared folders). However using this to redact material sounds a bit risky. Are there some use case where you could accept the potential mistakes (missing something that should have been redacted)?