Symmetry Systems Unveils AI Governance Tool
Symmetry Systems has launched Symmetry AIGuard, a platform designed to provide unified security and governance for AI systems. The product aims to offer visibility and control over external large language models, enterprise copilots, and internal AI services.
- The new AIGuard platform is built upon Symmetry Systems' existing DataGuard technology, which originated from a decade of DARPA-funded research at the University of Texas at Austin's Spark Lab. Co-founder and CEO Mohit Tiwari was a cybersecurity professor at the university, and the company was spun out of his research on information flow security. - Symmetry Systems has raised a total of $36 million over three funding rounds, with the most recent being a Series B round in August 2023. This funding is aimed at scaling its AI-powered Data Security Posture Management (DSPM) platform. - AIGuard provides a governance layer for what it calls "agentic AI," treating AI agents as first-class identities that require security and policy enforcement, similar to human users. The platform inventories all AI agents, maps their permissions and potential data "blast radius," and includes a sanctioning workflow to ensure no agent operates without authorization. - The platform's approach is data-centric, focusing on providing visibility into what data an organization has, who can access it, and how it is used, which supports a zero-trust security model. This is designed to help organizations with regulatory compliance for frameworks such as GDPR, HIPAA, and SOC2. - A key feature of Symmetry's technology is a comprehensive data classification system that uses over 400 sensitive data identifiers and 500 semantic data types to categorize information like PII, financial records, and legal filings. The company has announced plans to release this taxonomy as an open-source project to encourage industry-wide standardization. - AIGuard is designed to address four main pillars of AI security: managing shadow AI (unauthorized use of external LLMs), governing enterprise copilots, securing internal AI services, and managing the security of autonomous AI agents. This unified approach aims to provide a single control plane for security, legal, and data teams. - The technology is designed to be deployed across hybrid environments, including all major cloud platforms (AWS, GCP, Azure), SaaS applications, and on-premise databases, without data needing to leave the owner's control.
