White circle raises $11m seed for production ai control platform

At a glance:

• White Circle secured $11 million in seed funding to expand its AI‑model monitoring platform.
• Backers include senior figures from OpenAI, Anthropic, DeepMind, Hugging Face, Mistral, Datadog and Sentry.
• The service already processed over one billion API requests and counts Lovable and two major digital banks among its customers.

What the funding means for White Circle

White Circle announced an $11 million seed round on Tuesday, earmarked for product development and hiring across the United States, United Kingdom and Europe. The capital comes from a roster of personal investments by AI‑industry veterans: Romain Huet (OpenAI), Dirk Kingma (formerly OpenAI, now Anthropic), Guillaume Lample (Mistral), Thomas Wolf (Hugging Face), Olivier Pomel (Datadog), François Chollet (creator of Keras), Mehdi Ghissassi (formerly DeepMind), Paige Bailey (DeepMind) and David Cramer (Sentry). Their involvement underscores the platform’s relevance to the very labs that build the large language models White Circle is designed to police.

Platform overview and capabilities

Founded by engineer Denis Shilov—who went viral in early 2024 after a single prompt bypassed safety filters on every major AI model—White Circle offers a single‑API control layer that scans inputs and outputs in real time against customer‑defined policies. The service detects harmful content, catches hallucinations, blocks prompt‑injection attacks, flags model drift and identifies abusive users. Customers can configure custom enforcement actions such as rate‑limiting or outright bans, and they can feed labelled user feedback back into White Circle’s own models to improve accuracy over time.

Compliance, language support and certifications

The platform supports 150 languages and holds SOC 2 Type I and Type II certifications as well as HIPAA compliance, positioning it for use in regulated sectors like healthcare and finance. Elena Iumagulova heads design, ensuring the UI balances technical depth with usability for security and compliance teams.

Research contributions: CircleGuardBench and KillBench

Alongside the product launch, White Circle released two research papers. CircleGuardBench, published in May 2025, is a benchmark that evaluates how AI moderation models perform under real‑world conditions. KillBench ran more than one million experiments across 15 AI models from OpenAI, Google, Anthropic and xAI, uncovering preferences linked to nationality, religion, body type and even phone brand when models were asked to make decisions about human lives. The study also showed that structured‑output integrations—now standard in production AI deployments—caused refusal rates to collapse and biases to amplify.

Early traction and market positioning

White Circle’s platform has already served more than one billion API requests. Announced customers include Lovable and two of the world’s largest digital banks, though the banks are not named publicly. The company did not disclose its post‑money valuation or revenue, but the backing from high‑profile AI practitioners and the release of rigorous research suggest a strong market foothold in the emerging AI‑governance space.

Outlook and industry impact

Partner Ophelia Cai of Tiny VC praised the team’s “unusual combination of deep technical credibility and clear commercial instinct.” As AI models proliferate across hiring, healthcare, finance and security, White Circle aims to give enterprises a unified point of control without sacrificing security, compliance or risk. If the platform can scale its policy engine and maintain its compliance certifications, it could become a de‑facto standard for production‑grade AI governance.