SAMA Cybersecurity Framework

What is the SAMA Cybersecurity Framework?

The SAMA Cyber Security Framework (SAMA CSF) is a set of mandatory cybersecurity controls issued in May 2017 by the Saudi Central Bank (SAMA). It applies to the financial institutions SAMA regulates, called member organizations. It groups controls into four domains and expects each organization to reach at least maturity level 3 on a 0 to 5 scale.

For example, an insurer’s claims team starts using an AI agent that reads claim files, looks up the policy in a core system and drafts settlement letters. The AI agent sends customer names, national IDs and medical notes to a hosted model. When SAMA reviews the insurer, it will ask how this new data flow is governed, how the data is classified and protected, and how the model provider is managed as a third party.

The four domains

  • Cyber security leadership and governance. Strategy, policy, roles, awareness and the cybersecurity committee.
  • Cyber security risk management and compliance. Risk assessment, regulatory compliance and audits.
  • Cyber security operations and technology. Asset management, identity and access management, data and information protection, cryptography, logging and monitoring, and incident response.
  • Third party cyber security. Contracts and vendor management, outsourcing and cloud computing.

Maturity levels

Level Name What it means
0 Non-existent No documented control
1 Ad-hoc Controls exist but are inconsistent
2 Repeatable but informal Controls repeat but are not formally defined
3 Structured and formalized Controls are defined, approved and applied. This is the minimum SAMA expects
4 Managed and measurable Control performance is measured and reviewed
5 Adaptive Controls improve continuously based on results

Member organizations assess themselves against the framework, and SAMA reviews and audits those assessments.

SAMA CSF and the Saudi PDPL

The two often apply together but cover different ground. The SAMA CSF is a sector security framework: it asks whether a financial institution protects its systems and data with formal, measured controls. The Personal Data Protection Law (PDPL), enforced by the Saudi Data and Artificial Intelligence Authority (SDAIA), covers the rights of people whose personal data is processed and the rules for using and transferring it. A bank rolling out AI usually has to show both.

Why the SAMA CSF matters for AI systems

AI assistants and AI agents create data flows that older control maps never listed: prompts, retrieved context, tool calls, model outputs and logs. Under the framework, those flows need classification, access control, protection in transit and at rest, and monitoring. A hosted model provider is a third party, so outsourcing and cloud controls apply as well.

How Protecto helps

Protecto detects and classifies sensitive data in AI prompts, documents and tool calls, then replaces it with tokens before it reaches the model. Real values are revealed only to authorized users and tools under policy, and Protecto reports on how data policies are enforced, which gives audit teams evidence for review. See PDPL and SAMA compliance for AI.

Frequently asked questions

Who must comply with the SAMA Cybersecurity Framework?

Financial institutions regulated by the Saudi Central Bank, called member organizations. This includes banks, insurance and reinsurance companies, finance companies and other SAMA-licensed firms.

What maturity level does SAMA require?

SAMA expects member organizations to reach at least level 3, structured and formalized, on its scale of 0 to 5. Each level requires meeting all criteria of the levels below it.

Does the SAMA CSF cover personal data?

It covers data protection as part of security, including classification, cryptography and access control. The rights of individuals and the rules on using personal data come from the Saudi PDPL.

Related terms

Saudi PDPL · Data classification · Data encryption · Logs · Least privilege access

Book a demo

See Protecto on your own data

Book a demo to see how Protecto finds and masks sensitive data before it reaches any AI model.

New: Protecto now runs inside LiteLLM. Control what sensitive data your models and tools see.
Read the launch blog