New Security Layer for RadGenius Assistants

We just rolled out a new security feature for all RadGenius Assistants. The best part? You don’t have to lift a finger. It’s already included and running on your Assistants right now.

---

Why we added this

We care about keeping your Assistant’s private data safe. That means protecting your system instructions and knowledge-base files from people who might try to “hack” the Assistant into giving them away.

Until now, we encouraged you to add guardrails inside your Assistant’s instructions. But we’ve now taken some of that responsibility off your plate by adding a platform-wide security layer.

---

How it works

• Every single user message is checked for intent before it’s sent to your Assistant.
• If the message looks like it’s trying to pull out your instructions or knowledge-base, it gets blocked.
• The user sees a warning.
• You’ll get an email notification showing you what was flagged.
• The user can keep chatting, but anything suspicious gets stopped before your Assistant even starts answering.

---

Why this matters

• Stronger protection: Your proprietary information stays private.
• No extra cost: RadGenius covers the moderation, so you don’t pay for these checks.
• Save money: You can even remove some of the extra “don’t share this” reminders in your instructions, which means less text sent to the API.

---

What you need to do

Nothing. This is already live for all Assistants!