The AI security threat engineering leaders aren’t talking about.

https://zporigin.leaddev.com/technical-direction/your-tests-cant-catch-a-poisoned-model