PAIG is an open-source project for protecting generative-AI applications, covering security, safety, and observability. It provides a framework that lets point security and safety solutions be added to an application as the surrounding techniques change, without requiring the application to be rewritten.
Features
- Security controls: protection for generative-AI applications built into the surrounding framework
- Safety controls: integration points for safety solutions as approaches to writing generative-AI applications change
- Observability: visibility into how a protected application behaves
- Drop-in integration: point solutions plug in without rewriting the application they protect
