Saudi Arabia Is Letting an AI Adviser Into the Capital-Market Sandbox

Saudi Arabia has not licensed an AI financial adviser for the open market.

It has done something more specific.

On August 31, the Capital Market Authority said Lamha Tanbu had completed the commencement requirements for a FinTech experimental permit covering the use of artificial intelligence in advisory.

The permit itself dates to a CMA board decision on May 14, 2025. The new state change is that the company has now cleared the requirements to begin the experiment.

That distinction matters because regulatory sandboxes are designed to create controlled exceptions for testing. They are not a shortcut around the full securities rulebook.

"Allowed to test" is not the same as "fully authorized"

The CMA's FinTech Lab describes an experimental permit as a way for firms to deploy and test capital-market products with actual clients within specified parameters and timeframes.

The CMA public FAQ describes a two-year ExPermit, while the formal instructions say the experiment period may not exceed two years from the date business commences.

At the end, a company can execute an exit strategy or seek to deploy the product more widely. Wider deployment requires both the regulator and the applicant to be satisfied with the test outcomes, along with a strategy for operating at scale and the ability to comply fully with the Capital Market Law and its implementing regulations.

That framework is important context for the Lamha Tanbu announcement.

The CMA has created a specific activity category for "the use of A.I. in advisory," and its public list records Lamha Tanbu under that category with a permit date of May 14, 2025.

But the August announcement does not describe the exact model, investment universe, client limits, advice workflow or human-oversight design of the live test.

Those are still open questions.

AI advisory is not the same as robo-advisory

Saudi Arabia's sandbox already includes companies in robo-advisory.

The CMA lists "the use of A.I. in advisory" separately.

That wording suggests the regulator sees enough difference in the proposed technology or operating model to classify it independently, even if the customer experience may overlap with automated investment advice.

The distinction will matter more as AI systems move from portfolio allocation toward conversational recommendations, research generation, suitability analysis or agent-like actions.

A traditional robo-adviser can often be described through relatively fixed rules and portfolio construction logic. A generative or agentic AI adviser may produce less deterministic outputs and create different questions around explanations, consistency, data use and supervision.

The CMA announcement does not say which of those capabilities Lamha Tanbu will test, so it would be premature to assign them to the company.

But the new permit category creates a place for those questions to be tested under supervision rather than outside the regulatory perimeter.

The sandbox is designed to contain failure

The CMA's experimental-permit instructions are explicit that the FinTech Lab is not intended to create a risk-free environment.

Applicants are expected to identify significant risks, explain how those risks will be mitigated and maintain an exit strategy if the product is discontinued.

That is especially relevant for AI-based advice.

A test can generate useful evidence precisely because the product is not yet assumed to be ready for unrestricted deployment. The regulator can observe how the service behaves with clients, what controls fail, how disclosures work and whether the intended benefits justify a wider permission.

That is a different governance model from approving a technology based on a product description alone.

Saudi regulation is moving with the product, not after it

The Lamha Tanbu change is small compared with the multibillion-dollar AI infrastructure announcements surrounding LEAP 2026.

But it may be more informative about how Saudi Arabia wants AI to enter regulated markets.

The Kingdom is not treating "AI" as a separate legal universe. In capital markets, the technology is being brought inside an existing supervisory framework through a controlled test tied to securities activity.

That approach can support innovation without pretending that a new interface changes the underlying regulated activity.

If the system is giving investment advice, the fact that AI generates or supports the advice does not remove the investor-protection question.

The sandbox simply gives the regulator a way to examine that question before the product is allowed to scale.

For Lamha Tanbu, the next meaningful milestone will not be another AI announcement.

It will be evidence from the experiment.