Why AI pentesting deserves its own certification

## The AI threat surface is structurally different

Most pentest certifications were designed before tool-using agents existed. AIPentestX assumes the systems under test reason, plan, and call APIs.

## What we test for

- Prompt injection that bypasses tool guardrails
- RAG poisoning at ingestion time
- Agent abuse via misaligned tool-call schemas
- Output handler injection (XSS via LLM-generated markup)
- Supply-chain risk in published model weights

Every domain is exercised in the proctored exam *and* in the hands-on lab — you can't pass on theory alone.