White Circle raises $11m Seed for production AI control platform
May 12, 2026 – 7:43 am
Senior figures from OpenAI, Anthropic, DeepMind, Hugging Face, Mistral, Datadog and Sentry have backed the Seed. The platform claims more than one billion API requests served and lists Lovable and two of the world’s largest digital banks among its customers.
White Circle
A platform that helps companies monitor, secure and control AI models in production, White Circle has raised $11m in Seed funding, according to today’s announcement. The capital will support product development and hiring across the US, UK, and Europe.
The round includes investments from notable figures in the AI industry:
- Romain Huet (OpenAI)
- Dirk Kingma (formerly OpenAI, now at Anthropic)
- Guillaume Lample (Mistral)
- Thomas Wolf (Hugging Face)
- Olivier Pomel (Datadog)
- François Chollet (creator of Keras)
- Mehdi Ghissassi (formerly DeepMind)
- Paige Bailey (DeepMind)
- David Cramer (Sentry)
These investors represent senior practitioners from the labs that build the models White Circle is designed to monitor.
The Platform and Its Features
White Circle was founded by engineer Denis Shilov, who gained viral attention in 2024 after a single prompt bypassed the safety filters of major AI models, reaching 1.4 million views. This led to partnerships with Anthropic, OpenAI, and Hugging Face.
The platform offers:
- Real-time monitoring: Scans AI inputs and outputs against customer-defined policies.
- Harmful content detection: Identifies and flags harmful content, hallucinations, prompt injection attacks, model drift, and abusive users.
- Custom enforcement actions: Allows customers to set rate limiting and bans, incorporating user feedback to improve accuracy over time.
- Multi-language support: Supports 150 languages.
- Security and compliance: SOC 2 Type I and Type II certified and HIPAA-compliant.
Shilov’s Vision
As stated by Shilov, "AI is moving faster than our ability to guide it… We already trust it with decisions that touch millions of people… With White Circle, we’re finally giving companies everything they need to hold their AI accountable…"
Research Highlights
Two pieces of research have been released alongside the platform:
- CircleGuardBench: Tests AI moderation models’ performance in real-world conditions.
- KillBench: Analyzed 15 AI models from OpenAI, Google, Anthropic, and xAI, finding preferences linked to nationality, religion, body type, and phone brand when making life-or-death decisions. It also documented structured-output integers used by these models.