Introducing the Generative Application Firewall (GAF)
This addresses security challenges for developers and users of LLM applications by providing a unified defense mechanism, though it appears incremental as it builds on existing fragmented methods.
The paper tackles the problem of fragmented defenses in securing LLM applications by introducing the Generative Application Firewall (GAF), which unifies existing methods like prompt filters and guardrails into a single enforcement point, similar to a WAF for web traffic, and extends coverage to autonomous agents and tool interactions.
This paper introduces the Generative Application Firewall (GAF), a new architectural layer for securing LLM applications. Existing defenses -- prompt filters, guardrails, and data-masking -- remain fragmented; GAF unifies them into a single enforcement point, much like a WAF coordinates defenses for web traffic, while also covering autonomous agents and their tool interactions.