AI needs a safe place to run: sovereignty and the endpoint

Across Europe, enterprises are asking harder questions about how to adopt AI without giving up security, governance, or control.

AI agents are beginning to use credentials, open applications, access company data, and take action on an employee’s behalf. That changes the questions organizations need to answer.

It's no longer only, where is my data stored? But rather, where is the work happening? What can the agent reach? And who is in control when it acts?

AI needs a safe place to run: a secure environment where those controls can be enforced. Sovereignty is part of how organizations get there.

That was a central conversation this week at Omnissa ONE EMEA in Amsterdam.

Sovereignty is about control

As Europe seeks greater control and governance across the AI stack, sovereignty has to extend beyond where a model is hosted or where data is stored. It also has to cover where AI agents run, what they can reach and who controls the infrastructure beneath them.

That makes the endpoint (virtual or physical) the last mile of enterprise AI security and a critical enforcement point for governing how and where agents operate.

The endpoint is also where intelligence becomes action. The goal is to create a sovereign, secure environment that is safe for the agents working on your behalf and safe from agents that may be working against you.

The endpoint is becoming an AI control point

An agent can start a task with legitimate access and still behave in ways you didn't expect. It can change course while doing the work. It can move at machine speed. And one employee’s device could eventually have multiple agents operating with access to the same applications, data, and systems that person uses every day.

Traditional security models were built around people. Agentic AI requires us to think differently about endpoint security.

Detection and response remain important. But the first priority should be prevention: reducing the opportunity for an agent—malicious or otherwise—to exploit a vulnerability in the first place.

You cannot rely on the agent to police itself. The controls need to sit around it.

But that control doesn’t mean pumping the brakes on AI adoption. You shouldn’t have to choose between your AI ambition and control. The more responsibility we give AI, the more important accountability becomes.

Sovereignty is part of creating a safe place for AI

That's why today’s sovereignty announcement matters for AI.

At Omnissa ONE EMEA, we announced the expansion of the Omnissa Sovereign Solution (OSS) to six new markets across Europe and Asia, bringing OSS to eight markets in total, with seven additional markets planned later this year. OSS is our fully managed sovereign offering designed to give organizations greater control over where and how their digital workspace is managed.

The model keeps four things within the customer’s jurisdiction: customer data, service operations, technical support, and customer contract.

A certified local entity deploys and operates the service, support is delivered in-market, and the customer contracts with the local partner.

OSS is already being delivered in live markets through certified local partners, including GEMA International Hosting, the core OSS service provider partner for Omnissa.

Control also means choice. Across Omnissa Workspace ONE and Omnissa Horizon, customers can choose deployment models that align with their requirements, including operating within their own data centers.

New Windows Server OS support extends this sovereign management model beyond end-user devices to more of the enterprise technology environment.

Read the full announcement.

Why does this matter for AI? Because when agents act inside the enterprise, organizations need control over the environment where that action happens.

A safe place for AI to run brings together containment, sovereignty, and accountability: control over where agents operate, what they can reach, and the infrastructure beneath them. That gives enterprises a way to adopt AI securely and on their own terms.

For the agentic endpoint, that means prevention first: reducing the opportunity for any agent, malicious or otherwise, to exploit vulnerability, and remediating machine speed as attacks accelerate.

Why the agentic endpoint matters

Sovereignty is moving from a question of location to a question of control.

AI is moving from answering questions to taking action.

Those shifts meet at the agentic endpoint.

The era of the agentic endpoint has started, and Omnissa has spent 20 years securing the place where it's happening. We already manage devices and virtual workspaces inside some of the world's largest and most regulated enterprises, and we understand the device, its security posture, the applications on it, and the policies that govern it.

Now, we're here to help keep your enterprise safe from the agents that attack you, and safe for the agents that work for you, on every device, at any scale, on the platform you already trust. That's how you say yes to AI safely. The agents are new, but our commitment to the endpoint is not.