AI SECURITY • 330 / 397
Understand how untrusted text, tools and autonomy create new attack paths.
Input Guardrail
An input guardrail validates, filters or rejects unsuitable input before model processing.
Think of it like
Think of Input Guardrail as an input-trust or privilege problem: words can influence software decisions, so boundaries must be explicit.
Real life
Email, documents and web pages can become hostile inputs to AI systems through attacks involving Input Guardrail.
SRE lens
Block unsupported file types or obviously malicious control strings.
Remember thisAn input guardrail validates, filters or rejects unsuitable input before model processing.