1.5The OWASP LLM Top 10 at a Glance

The OWASP Top 10 for LLM Applications is the essential framework for AI security testing. Here is a high-level overview; Parts II and III of this book cover each item in depth with hands-on labs.

#	Vulnerability	Severity	What It Means
LLM01	Prompt Injection	Critical	Attacker manipulates the model via crafted input
LLM02	Insecure Output Handling	High	Model output used unsafely in downstream systems
LLM03	Training Data Poisoning	High	Malicious data in training corrupts model behavior
LLM04	Model Denial of Service	Medium	Crafted inputs exhaust resources or cause excessive costs
LLM05	Supply Chain Vulnerabilities	High	Compromised models, libraries, or data sources
LLM06	Sensitive Information Disclosure	Critical	Model reveals confidential data
LLM07	Insecure Plugin Design	High	Plugins/tools with excessive permissions
LLM08	Excessive Agency	High	Model takes actions without adequate oversight
LLM09	Overreliance	Medium	Blind trust in model output without verification
LLM10	Model Theft	Medium	Unauthorized extraction of model behavior or weights

Every entry in this table should have corresponding automated tests in your CI pipeline. This is the minimum security test coverage for any AI application.