Sovereignty and compliance

Data residency for AI in Saudi Arabia, the UAE and the GCC: PDPL, SAMA, NCA and in-Kingdom lakehouses

In Saudi Arabia, the PDPL is enforced by SDAIA, and SAMA and NCA expect banking and critical-sector data to stay in the Kingdom. In the UAE, the federal PDPL applies, and the Central Bank expects core banking and customer data to be processed in the country. For AI, residency covers the whole path: data, embeddings, prompts, model inference and logs. An in-country lakehouse with a local AI gateway keeps all of them inside the border.

Published . Last updated

On this page

Gulf governments and regulators are among the most explicit in the world about keeping data and AI in-country. That creates a practical problem for teams adopting AI: many data platforms and model APIs process data outside the country by default. This guide summarises the main rules in Saudi Arabia and the UAE, explains where AI breaks residency assumptions, and shows how an in-country lakehouse handles it.

What do Saudi rules require?

  • Personal Data Protection Law (PDPL): enforced by the Saudi Data and AI Authority (SDAIA) since September 2024, with restrictions on transferring personal data outside the Kingdom.
  • SAMA: financial institutions regulated by the Saudi Central Bank are expected to use infrastructure that meets in-Kingdom residency requirements.
  • NCA: the National Cybersecurity Authority's controls, including the Essential Cybersecurity Controls and cloud controls, apply to government and critical-infrastructure operators and address data localisation.

The practical reading for banks, telecom operators and government entities: regulated data, and the AI that processes it, should run in the Kingdom.

What do UAE rules require?

  • Federal PDPL (Federal Decree-Law No. 45 of 2021) governs personal data processing.
  • The Central Bank of the UAE expects core banking and customer data of financial institutions to be processed within the UAE.
  • Free zones such as DIFC and ADGM have their own data protection regimes.

Why does AI break data residency assumptions?

Because an AI feature moves data along a path, and every hop has a location:

HopCommon leak
Prompt assemblyCustomer data inserted into a prompt sent to a model abroad
EmbeddingsDocuments embedded by a cloud API in another country
InferenceThe model itself hosted outside the country
Logs and telemetryPrompts and answers logged in the vendor's home region
Assistant featuresSchemas, sample rows and query text sent to a wider region

Residency is set by architecture, not by the model vendor's marketing. If any hop leaves the country, regulated data has crossed a border.

How does an in-country lakehouse keep the whole AI path local?

DataNature runs the full stack in the country, control plane included:

  • Data in Apache Iceberg tables on in-country storage, ingested by Connect and StreamHouse.
  • Embeddings generated by DocuContext and stored in Milvus inside the deployment.
  • Inference through the AI gateway, pointed at a model served in-country: self-hosted on local GPUs, operator-managed, or an in-Kingdom provider.
  • Logs and audit in Atlas, inside the same deployment.
  • Governance: row filters, column masks and PII guardrails, with Arabic and English documents processed by the same pipeline.

Telecom operators often deploy in their own private cloud; banks in an in-country sovereign cloud or on-premises.

Which use cases matter most in the Gulf?

  • Telecom: churn, network analytics and agents for engineers, on data that must stay in-country.
  • Banking: customer 360, fraud on the StreamHouse hot path, AML network analysis with the knowledge graph.
  • Government: citizen-service agents grounded in official guidance, records digitisation.
  • Energy: asset maintenance and ESG reporting.

Frequently asked questions

Can DataNature run in Saudi Arabia?

Yes. DataNature can be deployed in-Kingdom on local infrastructure or on-premises, with the whole stack, including AI inference and logs, inside the Kingdom.

Does in-country residency rule out frontier models?

Not necessarily. Use in-country models for regulated data, and route non-sensitive workloads to other endpoints where your policy and legal advice allow. The AI gateway makes the routing explicit.

No. Requirements depend on your sector and licence. Confirm your obligations with qualified counsel in each jurisdiction.

Sources

How does DataNature Atlas keep this secure and compliant?

Every rule in this article is enforced in one place, inside your jurisdiction. DataNature Atlas is the platform's governance module:

  • Knows your data: a catalog with automated PII detection and classification across Bronze, Silver and Gold.
  • Controls access: role-based grants plus attribute-based policies, row filters and column masks (null, redact, partial, hash, year only).
  • Governs AI: Voir agents are scoped principals; masks, row filters and PII guardrails apply to every tool call and answer.
  • Proves it: lineage from source to metric to decision, and an audit log of who or what accessed which data.
  • Stays local: Atlas runs inside your deployment, so policies, lineage and logs never leave your jurisdiction.

These controls support GDPR, PIPEDA, Quebec Law 25, CCPA, HIPAA and sector rules such as OSFI E-23. Read DataNature Atlas in depth or see security and compliance.

See it on your own data.

Book a demo

Demos and quotes are handled by Dlytica, the company behind DataNature.