liveBook · Manning

Data sovereignty

As explained in LLM Customization and Fine-Tuning

The control of sensitive data within an organization's own infrastructure or jurisdiction so it does not leave the trusted boundary.

When sensitive data cannot be sent to a third-party API, data sovereignty becomes a deployment decision: keep inference within the organization’s infrastructure or an approved jurisdiction. Fine-tuning alone does not provide this protection; where the model runs determines whether data stays inside the trusted boundary. section 1.3

Common deployment choices include:

  • Run a model on-premises or in a private cloud. section 1.3
  • Use a managed service in the customer’s cloud region with data-residency and no-training commitments. section 1.3

A regulated deployment

A team can self-host Qwen3-4B with retrieval and leave its weights unchanged to meet a regulator’s boundary requirement. If the small model needs improvement to do the job, adaptation may help; it is not what keeps the data in-bounds. An in-house deployment may cost more at low volumes, yet remain usable when a frontier API is not. section 1.3 section 1.3

How LLM Customization and Fine-Tuning covers it

How deep it goes
Distinguishes deployment-location controls from adaptation and names hosting options, without detailing configuration.
Examples
Uses self-hosted Qwen3-4B with retrieval and regional managed-cloud inference as deployment examples.
What you can do after
Evaluate whether in-boundary deployment is necessary and whether adaptation can make its permitted model capable enough.
How it connects
Connects data sovereignty to fine-tuning, retrieval, self-hosting, private cloud, data residency, and third-party APIs.

Also explained in

Compare how 3 books explain data sovereignty →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of LLM Customization and Fine-Tuning. The section links point to the parts of the book it draws on.