AI Security

Taken two ways:

• Using AI to secure your product
• Using third party AI in your product securely

Stages

Visibility

Concerned with apps, agents, prompts, data stores, model, infrastructure

Security, Safety & Trust

Concerned with data leakage, prompt injection, denial of service, agent manipulation, data exfiltration, excessive agency, harmful content, AI Hallucinations

Supply Chain Risks

Concerned with model vulnerability, system prompt vulnerability