Gray Swan provides enterprise-grade AI security solutions to prevent data breaches, govern agent behavior, safeguard brand trust, detect failures, and harden AI systems. Their products include the Shade enterprise platform with Cygnal for continuous monitoring and adversarial testing, and the Arena for adversarial evaluation and AI red-teaming, plus AI red-teaming services. They also publish research and benchmarks such as AgentHarm to measure harms of LLM agents and collaborate with safety institutes to advance AI safety and security.
AI named AgentHarm in August 2026.
Brand page for grayswan-ai-2