# White Circle Secures $11 Million to Build the "Control Layer" for Enterprise AI Systems


Paris-based startup launches AI governance platform as enterprises grapple with hallucinations, data leaks, and prompt injection attacks in production AI deployments


White Circle, a Paris-based cybersecurity startup founded in 2025, has announced $11 million in seed funding to accelerate development of its AI control platform—a unified monitoring and governance solution designed to give security teams visibility and enforcement capabilities over AI models and agents in production environments.


The funding round, led by multiple angel investors, arrives at a critical moment: enterprises are rapidly deploying AI systems across customer-facing applications, internal operations, and critical workflows, yet most organizations lack adequate safeguards to detect when those systems malfunction, leak sensitive data, or are exploited by adversaries.


## The Threat Landscape: Why AI Governance Matters Now


Enterprise adoption of generative AI has accelerated faster than the security controls needed to govern it. A year ago, AI safety was largely an academic concern. Today, it's an operational necessity. Organizations face a converging set of risks:


  • Hallucinations and model drift: AI models produce confident-sounding but false outputs, misleading users or customers
  • Sensitive data leakage: Models inadvertently expose customer information, credentials, or proprietary data in responses
  • Prompt injection attacks: Adversaries craft malicious inputs to manipulate AI model behavior
  • Abuse and misuse: Users exploit AI systems for harassment, fraud, or policy violations
  • Compliance exposure: Unmonitored AI systems create blind spots for regulated industries (healthcare, finance, legal)

    • Until now, organizations have been deploying AI with essentially no runtime visibility—similar to running production databases without monitoring or logging. White Circle's platform attempts to fill that gap.


    ## What White Circle Does


    White Circle functions as a policy enforcement layer between AI applications and their users. The platform:


  • Monitors all AI inputs and outputs in real-time, scanning requests and responses
  • Applies organization-defined policies to detect policy violations, data leakage, and security risks
  • Identifies multiple threat classes including hallucinations, harmful content, prompt injection attacks, abusive users, and model degradation
  • Enforces actions such as blocking malicious requests, flagging or banning abusive users, and preventing agents from executing unauthorized actions
  • Supports 150 languages, enabling enforcement across global products and multi-language deployments
  • Improves over time through machine learning, with the company claiming its detection models become more accurate with use

    • The platform is built to work with any AI model or agent—not a specific vendor's systems—making it infrastructure-agnostic. This is critical because enterprises often use models from multiple providers (OpenAI, Anthropic, open-source models, proprietary internal systems).


    ### Key Capabilities


    | Capability | Benefit |

    |---|---|

    | Real-time input/output monitoring | Catch problems before they reach users |

    | Policy-based scanning | Enforce org-specific security rules |

    | Multi-language support | Global compliance and safety |

    | Sensitive data detection | Prevent customer PII and trade secrets leakage |

    | Agent control | Prevent AI agents from executing unauthorized actions |

    | Abuse detection | Identify and block malicious users |

    | Model degradation alerts | Flag when model performance declines |


    ## The Funding and Strategic Direction


    White Circle's $11 million seed round will fund three key initiatives:


    1. Product acceleration: Deepening the platform's detection capabilities, adding integrations with major AI platforms, and expanding language support

    2. Global hiring: Building engineering and go-to-market teams across the US, UK, and Europe

    3. Customer acquisition: Expanding from early customers to enterprise deployments


    "Until now there's not been a platform purpose-built to monitor AI's behavior, catch it when it goes wrong, or shape how it acts," said Denis Shilov, White Circle's founder and CEO. "With White Circle, we're finally giving companies everything they need to hold their AI accountable and optimize their models in a single place, without sacrificing security, compliance or risk."


    The funding reflects investor confidence in the AI safety market. White Circle joins a growing ecosystem of AI governance startups (including companies focused on model evaluation, red-teaming, and adversarial testing), but White Circle's focus on runtime enforcement and policy control sets it apart from competitors focused primarily on evaluation or testing.


    ## How It Works in Practice


    In a typical deployment, White Circle sits as an intermediary between an AI application and the underlying model. When a user submits a request:


    1. Inbound scanning: The user's prompt is analyzed for injection attacks, abuse, and policy violations

    2. Policy enforcement: Requests matching defined blocklists or risk thresholds are rejected or flagged

    3. Model execution: Approved requests proceed to the AI model

    4. Output analysis: The model's response is scanned for hallucinations, data leakage, and harmful content

    5. Action enforcement: Risky outputs are blocked, redacted, or logged; abusive users are flagged or banned

    6. Feedback loop: Detection results inform model improvement and policy refinement


    This approach is fundamentally different from traditional security controls, which typically focus on the infrastructure hosting the AI (network access, authentication, encryption). White Circle focuses on the behavior of the AI itself.


    ## Implications for Enterprise Security Teams


    For organizations deploying AI systems, White Circle's emergence signals an important market shift: AI safety is becoming a standard security control, not an optional research project.


    Security teams should consider several implications:


  • Governance is now a deployment requirement: Organizations that don't monitor AI model behavior risk regulatory penalties, customer data exposure, and reputational damage
  • Policy-driven enforcement is essential: As AI adoption scales, manual review becomes impossible; automated, policy-based controls are necessary
  • Multi-model environments need unified monitoring: With enterprises using models from multiple providers, a vendor-agnostic control layer is valuable
  • Compliance mandates are coming: Regulators in the EU, US, and elsewhere are beginning to require explainability and safety controls for AI systems; White Circle positions itself as a compliance enabler

    • ## Recommendations for Organizations


    Security and AI teams should:


    1. Inventory AI deployments: Map where AI is currently used (customer-facing, internal operations, third-party integrations)

    2. Define AI safety policies: Establish clear rules for acceptable model behavior, data handling, and user protection

    3. Evaluate monitoring solutions: Assess platforms like White Circle that provide runtime visibility and enforcement

    4. Test with non-critical systems first: Pilot AI safety controls on lower-risk deployments before rolling out enterprise-wide

    5. Integrate with security operations: Ensure AI safety alerts are routed to security teams and incident response workflows


    ---


    ## HackWire Analysis


    White Circle's $11 million funding validates a critical gap in enterprise AI security: most organizations have no visibility into how their deployed AI systems behave in production. This is not a theoretical problem.


    Real incidents demonstrate the urgency. In 2025, multiple organizations reported AI chatbots leaking customer data, apologizing for non-existent incidents, or insulting users. Some firms discovered their models were being manipulated through prompt injection attacks. Yet most organizations had no way to detect these failures in real-time—they only learned about them through customer complaints.


    White Circle addresses a legitimate market need. However, the broader pattern is important: the AI industry is moving faster than security can keep up. Every new capability (agents, function calling, reasoning models) creates new attack surfaces and failure modes. Defense-in-depth approaches that combine monitoring, policy enforcement, and architectural isolation will become essential.


    The startup also exemplifies a trend in security funding: governance and compliance are attracting capital at the same rate as breach-detection tools. Investors are betting that as AI regulation tightens (EU AI Act, sectoral mandates), demand for auditable, policy-compliant AI systems will drive significant revenue. White Circle benefits from this tailwind.


    One risk to monitor: policy-based safety systems can create false confidence. A well-monitored AI system is safer than an unmonitored one, but it is not infallible. Organizations should view White Circle (and similar tools) as part of a layered strategy—not a silver bullet. The safest AI systems are those that are also thoughtfully designed, regularly audited, and operated with human oversight.


    — HackWire Editorial


    ---


    ## Related Coverage


  • Read more in our [Breaches](https://www.hackwire.news/category/breaches) coverage
  • Cross-reference with [Vulnerabilities](https://www.hackwire.news/category/vulnerabilities) and [Malware](https://www.hackwire.news/category/malware)
  • Stay current via the [HackWire homepage](https://www.hackwire.news/)