Gray Swan AI

Gray Swan AI is a safety and security provider for the AI era, spun out of Carnegie Mellon University's AI safety research team. The company raised a $40M Series A and counts Anthropic, OpenAI, Meta, Google DeepMind and Snowflake among the organisations that trust it. It is scaling go-to-market operations and deepening partnerships with frontier AI labs, serving frontier AI labs, AI/ML enterprises and organisations deploying AI models and agents in production.

Its work centres on automated adversarial testing, continuous red teaming and runtime AI security - helping customers assess how their AI models and agents can behave in unintended ways. Three products make up the offering. Shade is automated adversarial red-teaming that finds AI vulnerabilities before attackers do. Cygnal provides real-time runtime defense, blocking adversarial inputs and unsafe outputs. Arena is described as the world's largest adversarial AI red teaming community, and generates the threat intelligence that powers both Shade and Cygnal.

The technical focus spans AI safety and security, AI model vulnerability assessment, AI agent safety, runtime threat detection and threat intelligence generation. The culture is research-rooted and technical, oriented towards the frontier labs and enterprise customers now adopting agentic AI systems.