Skip to content

Glossary: AI Security and Guardrails

Canary

Also written: canaries, canary-4417, leak canary, behavior canary, exfiltration canary

A made-up marker with no use except to show that something happened. The lab uses a secret marker in prompts (`canary-4417`), the harmless word PINEAPPLE, and the reserved `example.com` address.

Where the book teaches it

Related terms

ExfiltrationAttack success rate (ASR)