White Circle

White Circle provides a control layer platform for artificial intelligence systems in production, enabling companies to test, protect, observe, and improve their AI models. The company's technology is designed to secure AI applications by preventing unsafe inputs, jailbreaks, and prompt injections, while also offering tools to analyze user behavior and optimize model performance. The platform operates through a single API, where proprietary models monitor AI inputs and outputs in real-time based on custom policies set by the client. This allows for the detection of harmful content, hallucinations, model drift, and malicious user activity. The company was founded by engineer Denis Shilov, who gained significant attention in the AI community in 2024. Shilov discovered and published a universal jailbreak method that could bypass the safety filters of major AI models with a single prompt, exposing vulnerabilities that could lead to the generation of dangerous or illegal information. This viral discovery led to engagements with companies like Anthropic, OpenAI, and Hugging Face, and highlighted the market need for robust AI security and control, which prompted the creation of White Circle. The platform's design is led by Head of Design, Elena Iumagulova, with a focus on making AI oversight accessible to both technical and non-technical teams. In May 2026, the Paris-based startup, operating legally as Pumpkin Intelligence Inc., announced it had raised $11 million in a seed funding round. The investment came from a notable list of leaders and executives from prominent AI companies including OpenAI, Anthropic, DeepMind, Hugging Face, Mistral, and Datadog. White Circle serves a global customer base, which includes two of the world's largest digital banks, and has processed over one billion API requests. The company is SOC 2 Type I and II certified and HIPAA compliant, positioning it to serve enterprises in sensitive sectors like finance and healthcare. Keywords: AI control layer, AI safety, AI security, LLM security, AI observability, jailbreak prevention, prompt injection, AI model monitoring, AI governance, production AI, AI testing, AI optimization, model performance, user behavior analysis, real-time AI monitoring, Denis Shilov, AI guardrails, large language models, enterprise AI, AI risk management, AI compliance