Swordfish Security has released version 2.0 of its AI Security assessment framework. This methodology is designed to help cybersecurity teams evaluate the maturity of their AI implementation and is available for free.
The updated framework introduces two new domains: security for agentic systems and security for RAG pipelines. The agentic systems domain focuses on monitoring the behavior of AI agents that perform tasks without human intervention, while the RAG domain addresses data provenance and version control.
CEO Alexander Pinaev stated that the new version provides a practical roadmap for companies to identify current security gaps and prioritize future actions. The framework supports both proprietary models and those with open weights.
The methodology emphasizes continuous management of AI security through inventory tracking and incident logging. This approach is intended to mitigate risks associated with the non-deterministic nature of generative AI models.
