White Circle, founded in 2024 and based in Paris, builds a unified AI control layer providing safety testing, real-time security enforcement, evaluations and performance optimization for AI applications in production.
White Circle is a Paris-based AI infrastructure company founded in 2024 by Denis Shilov. The company builds a unified AI control layer covering safety testing, real-time security enforcement, evaluations and performance optimization for AI applications running in production. It raised $11m in seed funding and is backed by founders of OpenAI, Anthropic, HuggingFace, Mistral, DeepMind, Datadog and Sentry.
The White Circle Platform combines several capabilities. Automated red teaming is used to surface AI failures; custom guardrails provide low-latency enforcement at runtime; behavioural analytics offer clustering and custom metrics; and AI optimization covers dynamic prompts and intelligent model routing. The platform processes over one billion API requests annually and supports more than 150 languages, with unlimited custom policies and SOC 2 Type II and HIPAA compliance.
The company emerged from a viral 2024 demonstration by Shilov showing that a single prompt could bypass the safety filters of every major AI model. Its stated focus is on securing and controlling AI behaviour in production, at a point where AI systems increasingly make decisions affecting large numbers of users. Technical work spans AI safety, AI security, red teaming, guardrails, behavioural analytics, prompt engineering, model routing, LLM evaluation and performance optimization.
