Would homomorphic encryption or any other privacy enhancing technologies allow for a middle ground here?
Would allow you to scan the encrypted messages for incriminating or illegal content when there is reason to believe a specific account is doing something illegal (could require court approval, then a further process for full decrypted messages as part of court proceedings)
Ahh interesting. So you could not have one key encrypting the messages than a separate org-wide key encrypting the naughty word list. That makes sense. Thanks.
Would allow you to scan the encrypted messages for incriminating or illegal content when there is reason to believe a specific account is doing something illegal (could require court approval, then a further process for full decrypted messages as part of court proceedings)