From the course: Secure Generative AI and LLM Deployment
Join today to access over 25,600 courses taught by industry experts.
Securing sensitive inputs
From the course: Secure Generative AI and LLM Deployment
Securing sensitive inputs
One of the fastest ways an internal AI system gets risky is through what people paste into it. Not because they're careless, but because they're trying to get work done. They drop in customer records, they paste contracts, they include HR data, they add logs with secrets in them. From their point of view, they're just giving the system what it needs to help them. But from a security point of view, that's sensitive data entering a very complex pipeline. So the first step in securing sensitive inputs is accepting this. People will paste sensitive data unless you design around it. There are a few core strategies that actually help. First, reduce what people can paste in the first place. You can use client-side and server-side checks to look for things like credit card numbers, social security numbers, health identifiers, API keys, or passwords. When you detect them, you can block the prompt, mask the data, or warn the user before it it goes anywhere. Second, minimize what gets sent to…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.