CSAIChaptersEventsBlog
Learn how to uncover runtime risks, close governance gaps, and strengthen oversight. Register now for the June 9 webinar →

CSA Research

Best practices, guidance, frameworks and tools to help the industry secure the cloud. Read our research to get your questions around cloud security answered.
Research

CSA Research is created by the industry for the industry and is both vendor-neutral and consensus driven. Our research is created by subject matter experts who volunteer for our working groups. Each working group focuses on a unique topic or aspect of cloud security, from IoT, DevSecOps, Serverless and more, we have working groups for over 20 areas of cloud computing. You can view a list of all active research working groups. To find out more about how our research is created and the process we follow you can view the CSA Research Lifecycle.

Contribute to CSA Research

Peer reviews allow security professionals from around the world to collaborate on CSA research. Provide your feedback on the following documents in progress.

Publications in Review
Open Until

Latest Research

CSA Innovator Market Map

CSA Innovator Market Map

Release Date: 06/04/2026

The Agentic AI Security Innovator Market Map provides a curated view of the emerging vendor landscape focused on securing agentic AI systems. 

This market map identifies companies delivering capabilities that leverage CSA best practices across key areas of agentic AI security, including:
  • ...
RiskRubric v2 Concept Paper

RiskRubric v2 Concept Paper

Release Date: 06/04/2026

This CSA concept paper introduces RiskRubric v2, an evidence-based risk assessment framework designed to evaluate AI services across six pillars of trust, security, and operational integrity.

The paper examines:
  • How AI risk assessment must expand beyond models to include MCP servers and AI agents
  • ...
RiskRubric Scoring Methodology

RiskRubric Scoring Methodology

Release Date: 06/04/2026

The RiskRubric Scoring Methodology provides the technical foundation for evaluating and benchmarking the security posture of AI models, MCP servers, and AI agents. Designed to produce consistent, transparent, and reproducible risk scores, the methodology combines established risk management...