AI Adequacy Protocol: DIFC Seeks Feedback on AI Rules

Good rules work best when the people who follow them have a say. That is why it is encouraging to see the Dubai International Financial Centre (DIFC) open a public consultation on a draft framework called the AI Adequacy Protocol. If adopted, it would sort AI systems that handle personal data into three risk levels, each with its own certification rules. Better still, it would let firms use credentials they already hold from other regions, which could save time and effort.

What Is the AI Adequacy Protocol?

The AI Adequacy Protocol is a proposed framework for developing, deploying, buying and using AI systems that process personal data within the financial centre. It is still a draft, and the public consultation is a chance for companies and other interested parties to help shape how it will work. The consultation is an informal, non-legislative online survey that takes about 10 minutes.

The draft was prepared by a sub-group of an advisory committee that looks at AI adequacy. It builds on DIFC Regulation 10, which covers how autonomous and semi-autonomous AI systems process personal data, and on the DIFC Data Protection Law of 2020.

Three Risk Tiers, Explained Simply

The AI Adequacy Protocol would give each AI system a score from 1 to 3 on four factors: data volume, decision impact, autonomy, and data sensitivity. The total decides the risk tiers:

  • Standard: a score below 6
  • Enhanced: a score of 6 to 9
  • Critical: a score of 10 or more

Any system that makes legally significant automated decisions would also be classed as Critical. In simple terms, the more data, impact and independence a system has, the more care it needs.

What Each Tier Would Need

Under the draft AI Adequacy Protocol, requirements grow with risk:

  • Standard systems self-declare, then an accredited certification body assesses them, leading to a three-year certification
  • Enhanced and Critical systems need an Autonomous Systems Officer, a named person responsible for oversight
  • Critical systems also need quarterly monitoring metrics, plus board-level oversight for larger organisations

Credentials From Other Places Can Count

A particularly helpful idea is reciprocity. Companies with an EU CE marking, ISO 42001 or Singapore AI Verify can ask for reciprocity mapping against DIFC’s rules. An accredited certification body would identify any gaps, and a streamlined assessment would cover only the DIFC-specific parts.

The DIFC Commissioner would keep an equivalence registry covering the EU, UK, Singapore and selected US states. Each jurisdiction would be assessed against eight criteria and rated as high, moderate, or low compatibility with DIFC rules. For firms working in several markets, this could mean less repeated paperwork.

Extra Care for Sensitive Sectors

Some industries may have added requirements, such as:

  • Financial services: model risk management and fair lending analysis
  • Healthcare: clinical validation
  • Human resources: human review of rejections in recruitment

These examples show the draft protocol paying attention to the places where decisions affect people most.

Clear Responsibilities

The draft sets out duties for providers, operators, and deployers. Deployers carry ultimate accountability, and a deeming provision attributes a system’s actions to them. They must also keep visibility of their supply chain and hold model cards for third-party components. Clear responsibility helps everyone know who to turn to.

A Fair Process if Things Go Wrong

Enforcement under the protocol would run through five levels. It starts with guidance for minor documentation faults and goes up to revocation, penalties, and referral for prosecution in cases of harm or persistent violations. Appeals go first to an independent panel and then, if needed, to the DIFC Courts.

A Regulation 10 Accelerator is also planned, to test the new approaches over 6, 12 or 18 months. Testing ideas before they are final is a thoughtful way to learn.

The Background

DIFC’s Commissioner of Data Protection enacted the rules on 1 September 2023, and enforcement began in full in January 2026. DIFC says it aims to make its rules work smoothly with the growing set of international AI laws. Related documents include an accreditation and certification framework and an accelerator framework.

Why This Consultation Deserves Appreciation

The draft AI Adequacy Protocol deserves appreciation for several reasons:

  • It asks for feedback before the rules are final
  • It recognises trusted certifications from other regions
  • It matches the level of care to the level of risk
  • It gives clear roles, such as the Autonomous Systems Officer, and a fair route for appeals

Conclusion

The AI Adequacy Protocol is a thoughtful draft that aims to balance innovation with responsibility. By inviting feedback, recognising international credentials and setting clear duties, DIFC is helping companies prepare for the AI era with clarity.

If your organisation wants guidance on preparing for AI rules and responsible AI use, Horizontal8 is here to help. Reach out to Horizontal8 today, and let’s talk about your next step.

Frequently Asked Questions

What is the AI Adequacy Protocol? 

It is a draft DIFC framework for AI systems that process personal data within the financial centre.

How would AI systems be classified? 

By a score on data volume, decision impact, autonomy, and data sensitivity, giving Standard, Enhanced, or Critical.

Can existing certifications count? 

Yes. Companies with an EU CE marking, ISO 42001, or Singapore AI Verify can request reciprocity mapping.

What is an Autonomous Systems Officer? 

A person that Enhanced and Critical systems would need to appoint to oversee them.

How long does the consultation survey take? 

About 10 minutes, and it is non-legislative.

Free Brand Audit