AI Security & Governance Services Saudi Arabia
AI stopped being a tool your staff use and became a system that acts on your behalf, with credentials you issued. We govern what it can reach, test how it can be manipulated, and produce the evidence a regulator will ask for.
What usually starts the work
- AI tools are in use across the business and nobody has an inventory.
- An agent has been given API access and its permissions were never scoped.
- Personal data is being pasted into hosted models with no lawful basis recorded.
- A copilot reads internal documents and the retriever ignores source permissions.
- The board has asked who is accountable if an AI system gets something wrong.
What we assess and deliver
AI system inventory
Every AI system in use, including shadow AI found via expenses, SSO logs and extensions.
Impact & autonomy classification
Two axes: what happens if it is wrong, and can it act without a human.
Data layer controls
Lawful basis, retention and — critically — permissions enforced at retrieval time.
Agent permission review
Scoped, short-lived credentials per tool and human approval for irreversible actions.
Adversarial testing
Direct and indirect prompt injection planted in content the system actually reads.
Governance & evidence
Named owners, review cadence, and logging of prompt, context, tool calls and output.
Assess → Remediate → Secure → Assure
Assess
Understand your risk and compliance position.
Remediate
Close gaps with prioritized solutions.
Secure
Deploy the right technologies and controls.
Assure
Maintain continuous compliance and resilience.
Not sure where you stand? A short scoping call establishes what applies to you and what the work would involve. No obligation. Request an AI Security Review →
What applies in the Kingdom
| SDAIA | National AI direction, ethics principles and generative AI guidance — also the data protection regulator. |
| PDPL | Applies to personal data used in training, fine-tuning, retrieval or prompts, including cross-border model hosting. |
| NCA ECC | An AI system is still an IT system: inventory, access control, logging and third-party controls all apply. |
| ISO/IEC 42001 | AI management system structure that gives auditors a recognised framework. |
| EU AI Act | Reaches Saudi organisations when the system's output is used inside the EU. |
What you receive
- AI system inventory with owners
- Impact and autonomy classification
- Agent permission and tool-scope review
- Adversarial test findings with evidence
- Governance model and review cadence
- Regulatory mapping across SDAIA, PDPL and NCA
Related services and insights
Frequently asked questions
Is prompt injection something you can actually fix?
Not with a filter. Indirect injection — a payload inside a document or page the retriever pulls in — has no reliable detection. The defence is architectural: treat every model output as untrusted input, scope the tools an agent can call, and require human approval for anything irreversible. We test whether those boundaries hold rather than claiming the model is safe.
Does PDPL apply to data we put into an AI tool?
Yes. The law applies to personal data regardless of the technology processing it. Training, fine-tuning, retrieval and even pasting into a prompt all count, and using a model hosted outside the Kingdom is a cross-border transfer that must be treated as one.
Do we need ISO/IEC 42001?
It is not mandatory here. It is useful when you must demonstrate AI governance to a customer, regulator or board, because it gives auditors a recognised structure. Many organisations adopt the structure without pursuing certification and capture most of the practical benefit.
Where do most AI deployments actually fail an assessment?
On the ordinary controls, not the model. Missing inventory, a vector store that inherited none of the source system's permissions, standing credentials on an agent, and logging that captures the answer but not the retrieved context or tool calls. Those are NCA ECC failures as much as AI failures.
Ready to start?
Tell us where you are and we will tell you what the work actually involves — scope, sequence and timeline, before any commitment.
Request an AI Security Review