GCC / Middle East (GCC) / revised 28 Oct 2026

Saudi Arabia AI Data Sovereignty: PDPL and NDMO Rules

Saudi Arabia has two layers that matter for AI inference. The Personal Data Protection Law, administered by SDAIA, controls when personal data may leave the Kingdom. The NDMO's data management and personal data protection standards, together with national cybersecurity controls, govern how public bodies and their contractors classify and handle data, and push government and sensitive workloads toward in-Kingdom processing.

Test a sovereign route Talk to a sovereign architect

Citation: Personal Data Protection Law (Royal Decree M/19, as amended by Royal Decree M/148), Article 29 and the Regulation on Personal Data Transfer outside the Kingdom; NDMO Data Management and Personal Data Protection Standards
Regulator: Saudi Data and AI Authority (SDAIA), including the National Data Management Office (NDMO)
In force: 14 September 2023 (enforced from 14 September 2024)
Maximum penalty: Fines of up to SAR 5 million per violation, which may be doubled for repeat offences (Article 36); up to two years' imprisonment and SAR 3 million for disclosing sensitive data with intent to harm (Article 35)

Statutory text

The PDPL and its transfer regulation

The PDPL entered into force on 14 September 2023 after amendment in 2023, and controllers had a one-year grace period, so enforcement began on 14 September 2024. Article 29 permits transfers outside the Kingdom for specified purposes provided they do not prejudice national security or vital interests, offer an adequate level of protection, and are limited to the minimum data needed.

The Regulation on Personal Data Transfer outside the Kingdom, updated in 2024, sets out how that works in practice: transfers to jurisdictions SDAIA has assessed as adequate, and otherwise appropriate safeguards such as SDAIA's standard contractual clauses, binding common rules for corporate groups, or certificates of accreditation, with narrow exceptions. A risk assessment of the transfer is required in defined cases, including transfers that rely on safeguards or exceptions and continuous or large-scale transfers of sensitive data.

What the NDMO standards add

The NDMO, part of SDAIA, issued the Data Management and Personal Data Protection Standards and related national data governance policies. They require entities in scope to classify data (Top Secret, Secret, Confidential, Public), appoint data management roles, and apply controls proportionate to each class. For AI teams, classification is the hinge: a prompt drawn from a Confidential government dataset inherits that class and its handling rules.

Cybersecurity hosting expectations

Separately from privacy law, the National Cybersecurity Authority's control frameworks expect national organisations using cloud services to host and store their data within the Kingdom. Government entities and critical national infrastructure operators therefore treat in-Kingdom inference as the default, not an option.

Technical requirements for the inference path

01
Classification travels with the request
Carry the NDMO classification of the source data as a request attribute so the policy engine can apply the correct handling rule before routing.
02
In-Kingdom data plane for government tenants
Government and CNI tenants should resolve only to inference capacity physically located in Saudi Arabia, either a provider's Saudi region or self-hosted models in the customer's own environment.
03
Transfer risk assessment as route metadata
Where a private-sector route does leave the Kingdom on SDAIA SCCs, attach the required transfer risk assessment to the route so changes in destination force a review.
04
Detect Saudi identifiers in prompts
Scan for Saudi national ID and Iqama numbers, and for health and financial terms that make data sensitive under the PDPL, before any egress decision.

Prescribed YTT Secunet routing rules

#	Rule	Legal rationale
01	When X-YTT-Policy is saudi-pdpl-sovereign, deny all egress outside Saudi Arabia for requests classified Secret, Top Secret or Confidential.	NDMO classification and national cybersecurity hosting controls keep these classes in-Kingdom regardless of PDPL transfer tools.
02	For Public or private-sector personal data, permit foreign endpoints only with an SDAIA-recognised transfer tool referenced on the route.	PDPL Article 29 and the transfer regulation require adequacy or appropriate safeguards.
03	Mask Saudi national IDs and Iqama numbers before any permitted cross-border route.	Article 29 limits transfers to the minimum personal data necessary.
04	Block sensitive data on cross-border routes that lack a current transfer risk assessment.	The transfer regulation requires a risk assessment for continuous or large-scale transfers of sensitive data.

Policy pack excerpt

saudi-pdpl-sovereign

The rules above, written as Open Policy Agent Rego (v1 syntax). The gateway evaluates the pack in-process in every regional data plane before forwarding a request; YTT Secunet measures median evaluation under 3 ms in production.

Packs are versioned and signed. The version that decided each request is named in the residency receipt, so an auditor can match the decision to this exact text.

How the policy engine works

regosaudi-pdpl-sovereign.rego

package yttsecunet.policies.saudi_pdpl_sovereign

import rego.v1

restricted_classes := {"top_secret", "secret", "confidential"}

in_kingdom if input.endpoint.country == "SA"

has_transfer_tool if input.endpoint.transfer_tool in {"sdaia-adequacy", "sdaia-scc", "sdaia-bcr"}

deny contains "classified data must stay in-Kingdom" if {
	input.request.ndmo_class in restricted_classes
	not in_kingdom
}

deny contains "no PDPL transfer tool for foreign endpoint" if {
	not in_kingdom
	not has_transfer_tool
}

deny contains "sensitive data requires transfer risk assessment" if {
	not in_kingdom
	input.guardrail.sensitive
	not input.endpoint.transfer_risk_assessment_id
}

default allow := false

allow if count(deny) == 0

Audit checklist

Each item is verifiable against configuration, signed receipts or contract records. Tick through it with your DPO.

01Each dataset feeding AI prompts carries an NDMO classification recorded in the data catalogue.
02Requests from government or CNI tenants resolve only to endpoints located in Saudi Arabia, verified from receipts.
03Every cross-border route lists its PDPL transfer tool and the date it was established.
04A transfer risk assessment exists for each cross-border route carrying sensitive data.
05Saudi national ID and Iqama detection is enabled and appears in the masking log.
06The records of processing activities name each AI provider and destination country.
07Contracts with foreign model providers include SDAIA's standard contractual clauses where those are the transfer basis.

Saudi PDPL and NDMO standards: frequently asked questions

Is NDMO the privacy regulator?

SDAIA is the competent authority under the PDPL. The NDMO is the SDAIA office that issued national data management and personal data protection standards, which apply most directly to public entities and those handling their data.

Can a Saudi bank use a model hosted in Europe?

Possibly for some data, subject to the PDPL transfer regulation and any sector instructions from its regulator, but classified or regulated data is typically kept in-Kingdom. The answer depends on classification and the transfer tool, which is why the policy evaluates both.

When did enforcement start?

The PDPL entered into force on 14 September 2023 and became enforceable on 14 September 2024 after the one-year transition.

Sovereign route ready

Stop risking 4% global turnover GDPR fines on AI inference.

Deploy sovereign, cryptographically proven AI routing in 15 minutes.

Schedule an architecture demo Start 14-day trial