General Analysis

General Analysis is a security research lab based in San Francisco that focuses on adversarial simulations, evaluations, and runtime protection for AI systems, particularly autonomous agents and multi-agent workflows. The company addresses security challenges introduced by AI agents, which require broad access to tools, APIs, and data — creating conflicts with traditional least-privilege security models. Their core thesis is that standard governance-based and alignment-based approaches are insufficient to secure enterprise AI deployments, and that security must be active, adversarial, and continuous. Their primary offering is a security evaluation platform powered by proprietary adversarial agents. These agents analyze a customer's runtime environment, map the graph of tools and permissions, and generate multi-step exploit chains that target interactions between components. The platform is designed to surface system-level risks that may not be visible when evaluating individual components in isolation. The platform supports two complementary functions: - Red teaming: context-aware automated red-teaming that maps tool and permission graphs and generates multi-step exploit chains across interacting agents - Blue teaming: adversarial testing of security configurations, providing infrastructure for customers to experiment with and validate defenses The company also integrates with CI/CD pipelines to forecast vulnerabilities during the development and deployment process. Their team has backgrounds in cybersecurity, AI research, and quantitative trading. General Analysis is hiring research engineers, software engineers, and business roles, all based onsite in San Francisco.