# AI governance in Saudi Arabia: a practical guide

Which rules apply to an AI system in the Kingdom, who publishes each one, what is in force and what is still a draft, and where to start.

Octopus research team · 17 November 2026 · Governance & regulatory readiness, Data, platforms & sourcing

AI governance in Saudi Arabia rests today on the Personal Data Protection Law, NCA's cybersecurity controls, NDMO's data standards and SDAIA's AI Ethics Principles. SDAIA's Responsible AI Policy would add binding duties written for AI, but it is still a draft. Start with an inventory, a risk class for each system and a data map.

This guide lists each rule with its publisher, its status in November 2026 and what it asks of an AI system, then sets out the order in which to act. Every status comes from the publisher or from a dated legal briefing, and drafts are labelled as drafts. It is a practical guide to start from; your legal and security teams sign off.

## The rules that apply to an AI system in the Kingdom

**AI governance in Saudi Arabia: rule, publisher, status and what it asks, November 2026**

| Rule | Publisher | Status | What it asks of an AI system |
| --- | --- | --- | --- |
| Personal Data Protection Law (PDPL) and its Implementing Regulation | SDAIA (regulator) | In force since 14 September 2023, with one year to comply [1] [2] | Process only the personal data the use case needs. Transfer personal data outside the Kingdom only under the conditions and safeguards set out, with a risk assessment where required. Notify SDAIA within 72 hours where notice is required [3] [4] [5]. |
| Amendments to the PDPL Implementing Regulation | SDAIA | Draft. Public consultation ran from 6 October to 5 November 2026 [6] | Proposals include an impact assessment where emerging technologies are used or processing leads to automated decisions, and storing personal data inside the Kingdom while transfers stay possible [6] [7]. |
| Essential Cybersecurity Controls (ECC-2:2024) | National Cybersecurity Authority (NCA) | In force, 2024 edition [8] [9] | Applies to government agencies and their affiliated companies and entities, and to private owners and operators of critical national infrastructure. Four main domains and 108 main controls. Cloud and hosting controls bind in-scope entities that use cloud services [8]. |
| Cloud Cybersecurity Controls (CCC-2:2024) | NCA | In force, 2024 edition [10] | Controls for cloud service providers and for the entities that subscribe to their services. They matter when a model, its data or its logs run in the cloud [10]. |
| AI Ethics Principles | SDAIA | Published, first edition, September 2023 [11] | Seven principles for anyone who designs, develops, deploys or uses AI in the Kingdom. Four risk tiers. High-risk systems need conformity assessments before and after deployment, and unacceptable-risk systems are not allowed [11]. |
| Generative AI Guidelines | SDAIA | Published, first editions, January 2024 [12] [13] | Guidance on using generative AI tools safely: one edition for government staff, one for the public [12] [13]. |
| Responsible AI Policy | SDAIA | Draft. Public consultation ran from April to May 2026 [14] | Would make governance binding: four risk tiers with Critical Risk prohibited, registration of AI products, safety reports reviewed by SDAIA before high-risk deployment, and annual third-party audits of high-risk systems [15]. |
| Data Management and Personal Data Protection Standards | National Data Management Office (NDMO), SDAIA | Published, version 1.5, January 2021 [16] | Fifteen data management domains, including data classification and personal data protection. Data is classified as Top Secret, Secret, Restricted or Public [16] [17]. |

## Personal data: the PDPL

The PDPL came into force on 14 September 2023, and SDAIA gave organisations a year to bring their processing into line [1] [2]. It applies whenever an AI system processes personal data: names and ID numbers in scanned documents, employee records, customer messages and call transcripts.

Three decisions follow for any AI project. Purpose: the system processes only what the use case needs. Transfers: personal data leaves the Kingdom only under the conditions in the Implementing Regulation and the transfer regulation, with a risk assessment where required [3] [5]. Sending prompts or files that contain personal data to a model hosted abroad can count as a transfer. Breaches: the Implementing Regulation requires notifying SDAIA within 72 hours of becoming aware of a breach that could cause harm [3] [18].

SDAIA consulted on amendments to the Implementing Regulation from 6 October to 5 November 2026. The proposals would require an impact assessment where emerging technologies are used or where processing leads to automated decisions that affect people's rights or interests. They would also require controllers to store personal data inside the Kingdom, while transfers abroad remain permitted under the law [6] [7]. These are draft amendments, not adopted rules.

## Cybersecurity: NCA's ECC and CCC

The ECC apply to government agencies and their affiliated companies and entities, and to private sector entities that own, operate or host critical national infrastructure. NCA encourages every other entity in the Kingdom to use them as good practice [8]. For an AI project, the third-party and cloud controls usually matter most. A model provider, a hosting platform and an integration partner are all third parties.

One change in ECC-2:2024 matters for hosting decisions. The old control that hosting and storage must be inside the Kingdom was removed from the ECC. Data localisation controls moved to the National Data Management Office at SDAIA, and NCA tells entities to refer to NDMO before acting on localisation [8]. The Arabic text of the ECC is the reference version [9].

NCA has also consulted on AI Cybersecurity Guidelines. That consultation is closed. Check NCA's site for the final text before relying on it [19].

## AI ethics and risk: SDAIA's principles and the draft policy

SDAIA's AI Ethics Principles apply to everyone who designs, develops, deploys, uses or is affected by AI systems in the Kingdom, including public, private and non-profit entities [11]. There are seven principles: fairness; privacy and security; humanity; social and environmental benefits; reliability and safety; transparency and explainability; and accountability and responsibility.

The principles sort AI systems into four risk tiers. Systems with little or no risk, such as spam filters, carry no restrictions. Limited-risk systems are subject to the principles. High-risk systems need conformity assessments before and after deployment. Systems with unacceptable risk are not allowed [11]. The principles also ask each adopting entity to name the people responsible for AI ethics, led by the head of the entity or its chief data officer [11].

The draft Responsible AI Policy would turn this into binding rules. According to a November 2026 legal briefing, it would cover the public, private and non-profit sectors, roll out in phases starting with high-risk sectors and large companies, and use four tiers: Critical, High, Limited, and Minimal or No Risk. Critical-risk systems would be prohibited [15]. The Responsible AI Policy is still a draft; its public consultation ran from April to May 2026 [14].

How SDAIA guides adoption, as opposed to regulating it, is covered in [SDAIA's AI Adoption Framework, explained](https://pioctopus.ai/en/insights/sdaia-ai-adoption-framework).

## Data classification: NDMO

NDMO's standards set out fifteen data management domains, from data governance and data quality to data classification and personal data protection [16]. Classification has four levels: Top Secret, Secret, Restricted and Public [17]. For AI, classification comes first: it decides which data a model may see, and it shapes where the system can be hosted.

## Where to start: six steps in order

1. Inventory. List every AI system you run or buy, including AI inside vendor products, with a named person accountable for each.
2. Risk class. Classify each system on SDAIA's four tiers, and note where the draft policy's tiers would differ.
3. Data map. For each system, record which personal data it touches and why, and its NDMO classification.
4. Hosting decision. On-premises, a cloud provider in the Kingdom, or a transfer with its risk assessment. Check NCA's cloud controls and NDMO's position on localisation before you choose.
5. Oversight and records. Decide where a person reviews or overrides the system, and keep the logs that a breach notice or an audit would need.
6. Watch the drafts. Track the Responsible AI Policy and the PDPL amendments, and update the inventory when either is final.

The [Go-Live Gate Checklist](https://pioctopus.ai/en/toolkit/go-live-gate-checklist) turns these steps into the checks a system passes before it goes live.

## Aligned, not certified

PDPL, the NCA controls and SDAIA's principles belong to their publishers. A consultancy can design and review your AI systems so they are aligned with them. Octopus does not certify against them and claims no certificate from them.

This is the work of our practices in [AI governance, risk and regulatory readiness](https://pioctopus.ai/en/services/governance-risk-compliance) and [data, sovereign platforms and technology sourcing](https://pioctopus.ai/en/services/data-sovereign-platforms). The [AI governance and regulatory readiness review](https://pioctopus.ai/en/services#governance-review) takes 3 to 4 weeks with a fixed scope. It delivers an inventory of AI systems, each with a risk class; a gap map against PDPL, NCA ECC and CCC, SDAIA's AI Ethics Principles and the draft Responsible AI Policy; a draft AI office charter; and a 90-day remediation plan.

## Questions and answers

### Does Saudi Arabia have an AI law?

Not as one binding instrument written for AI. A November 2026 legal briefing describes AI governance in the Kingdom as resting largely on voluntary guidelines and non-binding principles [15]. AI systems are still covered by binding rules that are not AI-specific, such as the PDPL and, within their scope, NCA's controls [3] [8]. The draft Responsible AI Policy would add binding AI rules [14] [15].

### Does the PDPL apply to AI systems?

Yes, whenever the system processes personal data. The law has been in force since 14 September 2023 [1]. It governs the purpose of processing, transfers outside the Kingdom and breach notification, so it shapes what data a model may use and where it can run [3].

### Does using a model hosted outside the Kingdom count as a data transfer?

It can. If the prompts or files sent to the model contain personal data, the PDPL's transfer rules apply: transfers are allowed only under specific conditions and safeguards, with a risk assessment where required [3] [5]. Assess the transfer before design sign-off, because it often decides the hosting.

### Is SDAIA's Responsible AI Policy in force?

No. The Responsible AI Policy is still a draft; its public consultation ran from April to May 2026 [14]. A November 2026 legal briefing notes that it may change after the consultation [15]. Its direction is clear enough to act on now: an inventory, a risk class for each system and human oversight.

### Who must comply with NCA's Essential Cybersecurity Controls?

Government agencies in the Kingdom and their affiliated companies and entities, inside and outside the Kingdom, and private sector entities that own, operate or host critical national infrastructure. NCA encourages all other entities to use the controls as good practice [8].

## Related pages

- [AI governance and regulatory readiness review](https://pioctopus.ai/en/services#governance-review)
- [Go-Live Gate Checklist](https://pioctopus.ai/en/toolkit/go-live-gate-checklist)
- [PDPL, NCA and SDAIA: what they ask of AI](https://pioctopus.ai/en/insights/pdpl-nca-sdaia-in-practice)

## Sources

Figures are quoted as published. Forecasts and drafts are labelled.

1. [Saudi Press Agency, SDAIA announces the PDPL in force, September 2023](https://www.spa.gov.sa/14233f0780e)
2. [Saudi Press Agency, one-year period to comply with the PDPL](https://www.spa.gov.sa/N2136222)
3. [SDAIA, PDPL Implementing Regulation](https://sdaia.gov.sa/en/SDAIA/about/Documents/ImplementingRegulation.pdf)
4. [SDAIA, Standard Contractual Clauses for personal data transfer](https://sdaia.gov.sa/Documents/StandardContractualClausesForPersonalDataTransferEN.pdf)
5. [Umm Al-Qura gazette, Regulation on Personal Data Transfer outside the Kingdom](https://www.uqn.gov.sa/details?p=25412)
6. [Clyde & Co, SDAIA publishes draft amendments to the Implementing Regulations of the PDPL, 8 October 2026 (legal briefing; draft amendments)](https://www.clydeco.com/en/insights/2026/10/sdaia-publishes-draft-amendments-to-the-pdpl)
7. [Akin Gump, SDAIA proposes amendments to the PDPL Implementing Regulations: 10 key proposed changes, October 2026 (legal briefing; draft amendments)](https://www.akingump.com/en/insights/alerts/saudi-data-and-ai-authority-proposes-amendments-to-implementing-regulations-of-the-pdpl-10-key-proposed-changes)
8. [NCA, Essential Cybersecurity Controls ECC-2:2024 (English), including Appendix C on data localisation](https://cdn.nca.gov.sa/api/files/public/upload/86e09090-44e4-481f-bc28-355673607654_ECC--2024-EN.pdf)
9. [NCA, Essential Cybersecurity Controls ECC-2:2024 (Arabic, the reference text)](https://cdn.nca.gov.sa/api/public/cms/files/e4c81aa9-a609-4823-abab-347283e29731_ECC-2-2024-AR-19092024-.pdf)
10. [NCA, Cloud Cybersecurity Controls CCC-2:2024](https://cdn.nca.gov.sa/api/files/public/upload/6d5408a3-d8e6-4e96-963b-2c7198e5b7c2_CCC-2-2024-EN-.pdf)
11. [SDAIA, AI Ethics Principles, first edition, September 2023](https://sdaia.gov.sa/en/SDAIA/about/Documents/ai-principles.pdf)
12. [SDAIA, Generative AI Guidelines for Government, first edition, January 2024](https://sdaia.gov.sa/en/SDAIA/about/Files/GenAIGuidelinesForGovernmentENCompressed.pdf)
13. [SDAIA, Generative AI Guidelines for the Public, first edition, January 2024](https://sdaia.gov.sa/en/SDAIA/about/Files/GenerativeAIPublicEN.pdf)
14. [Saudi Press Agency, SDAIA invites the public and entities to comment on the draft Responsible AI Policy, April 2026](https://www.spa.gov.sa/en/N2551533)
15. [AS&H Clifford Chance, briefing on SDAIA's draft Responsible AI Policy, 4 November 2026 (legal briefing; the policy is a draft)](https://www.cliffordchance.com/content/dam/cliffordchance/briefings/2026/11/responsible-ai-policy-shift-guiding-regime-supervisory-one.pdf)
16. [NDMO, National Data Management and Personal Data Protection Standards, version 1.5, January 2021 (copy on Qanoniah)](https://qanoniah.com/File/2k3ZM8gW5oZkVN2LRDEebGwYj)
17. [Ministry of Education, Data Governance Policies, version 1.1 (NDMO policies on data classification)](https://moe.gov.sa/media/uploads/61/616eefd6da4f4ea4c4790a5f700a8005523cfad70d06f3ecec25057215846285.pdf)
18. [SDAIA, personal data breach notification service](https://dgp.sdaia.gov.sa/wps/portal/pdp/services/personaldatabreachnotification)
19. [NCA, public consultations, including the AI Cybersecurity Guidelines (closed)](https://nca.gov.sa/ar/public-consultations/ai-cybersecurity-guidelines/)

## Related practices

- [AI governance, risk and regulatory readiness](https://pioctopus.ai/en/services/governance-risk-compliance)
- [Data, sovereign platforms and technology sourcing](https://pioctopus.ai/en/services/data-sovereign-platforms)

## Put it to work.

The tool that goes with what you just read.

[Go-Live Gate Checklist](https://pioctopus.ai/en/toolkit/go-live-gate-checklist): The conditions an AI system should meet before it enters live operations.

## Keep reading

- [PDPL, NCA and SDAIA: what they ask of an AI project](https://pioctopus.ai/en/insights/pdpl-nca-sdaia-in-practice): Compliance starts with the design. PDPL, NCA's Essential Cybersecurity Controls and SDAIA's AI Ethics Principles shape almost every AI project in the Kingdom. Here is what each asks of you in practice.
- [How a Saudi government entity starts with AI](https://pioctopus.ai/en/insights/ai-adoption-government-entities): Start from SDAIA's national references, set up an AI office with a named person accountable, measure readiness in four weeks and choose the first three use cases.
- [SDAIA's National AI Index: what it measures and how to prepare](https://pioctopus.ai/en/insights/national-ai-index-saudi-arabia): The National AI Index scores government entities on three pillars, seven sub-pillars and 23 domains, with evidence behind every answer. The structure, the six maturity levels, the measurement cycle and the evidence to keep.

## Want to talk about your own operations?

- [Book a session](https://pioctopus.ai/en/contact)
- [Get the checklist](https://pioctopus.ai/en/toolkit/go-live-gate-checklist)

---
Web page: https://pioctopus.ai/en/insights/ai-governance-saudi-arabia
Arabic version: https://pioctopus.ai/insights/ai-governance-saudi-arabia.md
Index for AI agents: https://pioctopus.ai/llms.txt
