This AI Startup’s Army Of 15,000 Hackers Pressure Tests Claude, GPT-5 And Gemini

Gray Swan, a Pittsburgh AI security startup, raised $40 million in Series A funding at a $200 million valuation.

Its Arena platform uses about 15,000 security professionals to red-team models like Claude, GPT-5, and Gemini for jailbreaks, prompt injection, and harmful behavior.

Major AI labs including OpenAI, Anthropic, Google DeepMind, Meta, xAI, and ByteDance have used the platform, which is cited in 11 frontier model system cards.

The company plans to expand beyond model labs into enterprise AI security as demand grows for pre-deployment safety testing.