OmniAI reposted this
New privacy features coming to OmniAI this week! 🔐 Here's a peek at the secret sauce. If you're in a regulated industry (AI applications or not) PII redaction is going be best practice. But simple redaction isn't going to cut it for LLM applications. The models need context to function correctly. i.e. replacing a user's name with "**** *****" is going to produce some unpredictable results. Instead we create a lookup table that substitutes the PII for a semantically similar value. That means `Dan Smith` is substituted for a randomly generated `John Baker` at query time. And the real user information is stored in a secure fashion. And the same substitution is applied as the data comes back from the model. Because we know `John` is a substitution for `Dan`, we're able to swap out response values so the user gets the expected output.