COAE is a 7-day AI red-team exam built with Google (aligned to SAIF): you assess an AI-driven environment (an AI app, an MCP server and separate ML challenges), then submit a commercial-grade report. It covers adversarial ML, prompt injection, LLM output exploitation, AI app/system security, privacy and defense.
<aside>
📝 Your methodology
Each phase is a blank you fill in. Rename Fill in to a step in your own workflow (renaming updates links), and add or delete phases as you need.
</aside>