White Circle
A seed-stage Paris real-time AI control layer endorsed by senior figures from OpenAI, Anthropic, Mistral, and DeepMind.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
White Circle, founded in 2025 and based in Paris, offers a real-time control layer for AI applications that monitors model inputs and outputs to catch prompt injection, data leakage, hallucinations, and misuse. It unifies AI safety evaluation and real-time enforcement and publishes CircleGuardBench, its own benchmark for AI moderation model performance; it is SOC 2 and HIPAA compliant across 150-plus languages.
It is seed-stage and has been endorsed by senior figures from OpenAI, Anthropic, Mistral, and DeepMind.
Innovation Matrix Assessment
Raised $11M seed (announced 2026-05-12) within roughly a year of founding, from individual investors at OpenAI, Anthropic, Mistral, Hugging Face, and DeepMind.
Company is very early stage (founded 2025); public case studies or named enterprise deployments were not found in coverage reviewed.
$11M seed closed 2026-05-12 drew unusually senior individual backers across major AI labs, a notable signal of industry credibility.
Combines safety evaluation and real-time enforcement in a single control layer, plus a self-published benchmark (CircleGuardBench), differentiating from point AI-guardrail tools.
Benchmark and compliance claims are self-published; no independent third-party validation was located.
AI governance and runtime control of model behavior is a rapidly growing CISO concern as enterprises deploy agentic AI in production.
Why CISOs Should Care
White Circle offers a real-time control layer for AI applications that monitors model inputs/outputs to catch prompt injection, data leakage, hallucinations, and misuse, giving security teams a way to govern AI agent behavior in production.
What Makes It Different
Unifies AI safety evaluation and real-time enforcement in one platform, built with CircleGuardBench, its own published benchmark for AI moderation model performance, and is SOC 2/HIPAA compliant across 150+ languages.
The Matrix Verdict
45/100 — EMERGING / UNRANKED
Very early-stage but notably endorsed by senior figures from OpenAI, Anthropic, Mistral, and DeepMind, suggesting credible technical positioning in the fast-moving AI governance/control space.
Editorial Note: Claims vs. Verified Findings
Seed round and investor list independently verified via SecurityWeek, Fortune, and Tech.eu. SOC 2/HIPAA compliance and CircleGuardBench benchmark results are vendor-published claims.
Sources
Alternatives to White Circle
Adaptive Security
AI-driven platform that simulates deepfake, voice, and multichannel social-engineering attacks to train and test organizations against next-generation phishing.
Quilr
CISO ReviewedEarly-stage agentic AI security startup building a 'Service-as-Software' platform to guard against human-related breaches and secure AI agent…
Zenity
Governance and security platform for AI agents and low-code/no-code development, securing agent identity, permissions and behavior across the…
Tenzai
An agentic AI penetration testing startup building autonomous 'AI hackers' to find and validate exploitable vulnerabilities at a…
Gray Swan AI
Carnegie Mellon-linked startup building automated red-teaming and runtime guardrails that frontier AI labs use to stress-test their models…
Alice (formerly ActiveFence)
Israeli AI security company (rebranded from ActiveFence in January 2026) offering a lifecycle platform to test, guard, and…