Open-source AI Needs Security: Why Cyberhaven Is Joining the Open Secure AI Alliance

By Volodymyr Kuznetsov, Chief Technology Officer and Co-Founder, Cyberhaven

Today, Cyberhaven is joining the Open Secure AI Alliance to help advance a future in which enterprises can adopt open-source AI without compromising security, compliance, or control over their data.

Cyberhaven is betting on a world where companies are free to choose from many models, agent frameworks, and open harnesses, as well as run them on infrastructure they control.

That choice matters.

Enterprises will not standardize on a single model or AI platform. They will use frontier models, open models, specialized models, and increasingly autonomous agents across endpoints, developer environments, cloud infrastructure, and business applications.

The AI stack will continue to change rapidly. The security of enterprise data cannot depend on which model or harness happens to be in use.

That is why Cyberhaven built its AI Security product, debuting at Black Hat this week, to be model- and harness-agnostic from the ground up.

Our goal is to make broader AI adoption possible.

Security Must Follow the Data

Traditional security tools often focus on a single transaction: a prompt, an upload, an API call, or a network connection.

But AI risk rarely exists in a single transaction.

A seemingly harmless prompt may contain source code copied from a private repository, customer information extracted from a business application, or intellectual property assembled across multiple documents. An agent may then transform that information, invoke additional tools, and send the output somewhere entirely new.

To secure these workflows, organizations need more than prompt inspection.

They need to understand:

• What sensitive data an AI system or agent accessed
• Where that data originated
• Who or what interacted with it
• How it was transformed
• Where it moved
• Whether its use violated policy or regulatory requirements

Cyberhaven reconstructs that complete chain of custody.

This data-centric context allows security teams to apply consistent controls across open and proprietary models, different agent frameworks, and infrastructure operated in the cloud, on premises, or at the endpoint.

The model can change.
The harness can change.
The infrastructure can change.
The organization’s data-security policies should not.

Securing the Full Agent Stack

AI agents expand the attack surface because they do more than generate content. They access files, invoke tools, query systems, execute actions, and communicate with other agents.

That creates a fundamental security requirement: enterprises must understand not only what an agent is doing, but what data it is doing it with.

Cyberhaven extends visibility and enforcement across the full AI workflow, helping organizations identify risky data access, investigate agent activity, and prevent sensitive information from being exposed or misused.

This is the security foundation enterprises need to confidently adopt open models, harnesses, and more.

Open Innovation Requires Security That Protects the Full Lifecycle of Data

Open-source AI and secure AI are not opposing ideas.

Openness can strengthen security by giving organizations and defenders greater transparency, flexibility, and control. But that openness must be matched with strong identity, isolation, evaluation, runtime visibility, and data protection.

By joining the Open Secure AI Alliance, Cyberhaven will work alongside NVIDIA and other organizations across AI, infrastructure, cybersecurity, and open source to help build those protections.

The future of enterprise AI should give companies the freedom to choose the best models, harnesses, and infrastructure for their needs.

Cyberhaven is making sure they can do so without losing control of their most valuable data.