Data Residency Was The Easy Part

Direct Source Verification: This story is aggregated from Forbes (forbes.com). Full reporting rights and copyright belong to the primary publisher.
Nick Burling, Chief Product Officer at Nasuni.

Nick Burling, Chief Product Officer at Nasuni.

getty​A company can keep every sensitive file within the required geographic boundaries and still lose control over where its information goes. An AI application processing that information through a service operating in another jurisdiction may be enough to introduce sovereignty risks.

Enterprise data sovereignty, or an organization’s control over how its data is used and which laws govern it, has traditionally been approached through data residency policies and regional infrastructure. But those safeguards can only go so far when applications process that information across national borders, potentially subjecting it to different legal requirements.

When an AI workload processes enterprise information, that data may pass through services operating beyond the jurisdiction where the original file resides. The source data can remain within its required geographic boundaries while information derived from it is processed elsewhere.

If an enterprise wants to scale AI without compromising sovereignty, it must be able to deploy workloads wherever they deliver the most value while maintaining demonstrable control over the information those workloads use.

The difficulty is that even routine AI use cases can introduce jurisdictional exposure. A financial services organization might store customer records in Europe to meet data residency requirements, then use an AI assistant to analyze them. While the original files remain in Europe, the assistant could send information to a model operating in another jurisdiction, raising sovereignty concerns even though the source data remains exactly where it belongs.

These secondary data artifacts deserve greater scrutiny because they can contain sensitive information even when they appear removed from the original documents. They may also persist in logs long after an AI interaction ends, creating additional exposure that organizations need to account for.

Under the EU’s General Data Protection Regulation, transfers of personal data outside the European Economic Area are subject to specific conditions and safeguards. Meanwhile, the U.S. CLOUD Act allows authorities, through qualifying legal processes, to require providers subject to U.S. jurisdiction to produce data within their custody or control, regardless of its storage location.

These laws confirm why the location of an AI provider’s infrastructure tells only part of the story. The provider’s legal jurisdiction also determines which authorities may compel access to the information it handles.

Keeping every AI workload within the same jurisdiction as its source data may seem like the safest approach. For a hospital processing sensitive patient information, local execution may be necessary. A manufacturer analyzing internal production data, however, could have greater flexibility in using specialized models from external providers.

Applying identical restrictions across these use cases can unnecessarily limit AI adoption. Organizations should be able to choose where each workload runs based on its regulatory requirements while retaining the flexibility to use specialized AI capabilities where appropriate. That may mean keeping sensitive workloads within private infrastructure and allowing others to use external services under established safeguards.

Hybrid architectures provide that flexibility, provided governance extends across the environments involved. This becomes especially important as organizations move from AI assistants toward autonomous agents.

Autonomous AI agents make these decisions more complex. Unlike applications that respond to individual user requests, agents can independently access information and interact with services across different environments. Each interaction can introduce another jurisdiction into the workflow, making it harder for organizations to maintain consistent control over sensitive data.

The practical priority is ensuring that sovereignty policies follow information throughout an AI workflow. That starts with understanding the information an application can access and the restrictions attached to it. A document containing regulated personal information may require different processing controls from an internal knowledge article. Those distinctions need to inform which services can access the material and where processing can occur.

Those policies must extend to the information AI generates, which can carry the same sensitivity as the original documents. Organizations need to control how that information is retained and who can access it, with visibility into the providers handling it. Sensitive portions of a workflow can then remain within approved infrastructure while other functions run elsewhere.

Finally, organizations need evidence that these controls are working. An audit trail that records who accessed a source file provides limited visibility into what happened afterward. AI governance increasingly requires tracing subsequent processing steps, including which service handled the information and, where relevant, where outputs were retained.

This level of accountability can also support adoption. When enterprises can demonstrate that an AI workload operates within established boundaries, they have a stronger basis for approving it for production use.

A company may have complete records of where its sensitive files are stored and still struggle to explain what happened to that information once AI accessed it. Without that visibility, sovereignty becomes difficult to enforce and even harder to prove.

You can’t prove sovereignty for what you can’t trace. Organizations that can provide that evidence will have greater freedom to adopt AI without compromising control.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Original Source
https://www.forbes.com/councils/forbestechcouncil/2026/11/18/data-residency-was-the-easy-part/
Visit Forbes ↗
SHARE STORY:
𝕏 f in

Related Coverage in Health