Back to Blog
AI Integration

UAE AI regulation in 2026: the rules that actually apply to your business

7 min read
Founders
CTOs
Compliance Leaders

Is there an AI law in the UAE in 2026?

No. As of October 2026 the UAE has no single AI law. Businesses follow a patchwork: DIFC Regulation 10 for personal data processed by AI systems in DIFC, the Central Bank's February 2026 guidance note for licensed financial institutions, and the federal data protection law. A new Federal Authority for AI and Data can propose legislation.

General guidance, not legal advice.

Dubai AI Campus has just doubled its targets

On 26 October 2026, at the Dubai AI Festival, Sheikh Hamdan bin Mohammed approved new targets for Dubai AI Campus: more than 1,000 AI companies and over 5,000 highly skilled jobs by 2030. According to the Dubai Media Office, the campus has attracted more than 500 AI companies since it launched in May 2024. Those companies have raised over AED 1 billion from private equity and venture capital funds, and the campus has helped create more than 2,000 specialised tech jobs. It passed its original five-year target for companies and investment two years early.

That growth means more businesses are building AI into their products, and more of them are asking the same question: which rules apply to us?

Is there a single AI law in the UAE?

Not yet. As People of Internet's analysis puts it, the UAE has no horizontal AI law. Instead, AI is governed in layers: free-zone data rules, sector guidance from regulators and a new federal authority. Which layer applies to you depends on where your entity is licensed, what sector you're in and whether your AI touches personal data.

DIFC Regulation 10: AI systems that process personal data

If your business is in DIFC, the key rule is Regulation 10 of the DIFC Data Protection Regulations. It covers machine-based systems that work in an autonomous or semi-autonomous way, process personal data and generate an output. DIFC's guidance says purely deterministic automation, fully controlled by humans, isn't meant to be caught by it.

The main points:

  • Who is responsible. The business that deploys the system is treated as the controller, and a provider operating it on that business's behalf is treated as the processor.
  • Notice. Users must get clear and explicit notice when they first use the system, including what it does with personal data, what it produces and how that output is used.
  • Human-defined purposes. A system may only process personal data for purposes that are human-defined or human-approved, or within limits that humans have set.
  • Design principles. Systems must be designed to be ethical, fair, transparent, secure and accountable.
  • Human intervention. On request, the business must be able to show how the system seeks human intervention where processing could have an unfair or discriminatory impact on someone.
  • High-risk processing. Using a system for high-risk processing requires certification requirements set by the DIFC Commissioner of Data Protection, and an appointed Autonomous Systems Officer. The regulation describes a certification-based approach rather than licensing, with the details to come in future guidance, so check the current status with the Commissioner before you plan a high-risk use.
  • Financial services: the Central Bank's AI guidance note

    If you're a licensed financial institution, including an insurance provider, the Central Bank of the UAE's guidance note on AI and machine learning, issued on 11 February 2026, is the reference to know. It sets out flexible principles rather than fixed rules, and the Central Bank expects institutions to use it when writing their own AI policies.

    What it covers in practice:

  • Governance. Boards and senior management are accountable for AI systems, and institutions should keep an inventory of every AI model they use, with at least its name, purpose and risk rating.
  • Fairness. AI should be tested for bias at least once a year, and whenever a model is upgraded, materially changed or replaced.
  • Transparency. Customers should be told when AI is used, especially for decisions that affect their access to products such as loans or insurance claims, and when they're talking to an AI application. Disclosures should be in plain Arabic and English.
  • Human oversight. AI should operate under meaningful human oversight, matched to the risk. Customers should be able to ask for a human review or explanation of an AI decision, and institutions should always be able to stop an AI system through human intervention.
  • Outsourcing. Institutions stay responsible for AI functions they outsource.
  • Mainland businesses: the federal data protection law

    On the mainland, the main rule is the Personal Data Protection Law (Federal Decree-Law No. 45 of 2021). According to a June 2026 legal analysis, the law is in force but its implementing regulations, which cover areas such as legal basis, data subject rights, cross-border transfers and breach notification, had still not been issued, and no supervisory authority had been clearly named for private sector oversight. The DIFC and ADGM data protection regimes, by contrast, already have active enforcement.

    The new Federal Authority for AI and Data

    On 14 June 2026, Sheikh Mohammed bin Rashid announced the Federal Authority for Artificial Intelligence and Data. According to the same analysis, it reports to the Cabinet, is led by the Minister of State for Artificial Intelligence, Digital Economy and Remote Work Applications, and brings together the UAE's AI Office, the TDRA's Information and Digital Government Sector and the Emirates Data Office. Its mandate includes setting national AI and data policy, proposing legislation and strategies, and setting standards. How quickly it moves on data protection enforcement is still an open question, so it's worth watching.

    How we build AI features that hold up under any of these rules

    Whichever layer applies to you, the same habits make an AI feature easier to defend. This is how we work on AI projects:

  • Tell users. Make it clear when AI is involved, what it does with their data and what it produces.
  • Keep a human in the loop on decisions that matter. Build in review steps for anything that affects someone's money, access or rights, and a way to switch the system off.
  • Know where data goes. Map what personal data the system uses, where it's stored and which third parties see it.
  • Log what the system does. Keep an audit trail of inputs, outputs and model versions, so you can explain a decision later.
  • Test for bias and drift. Check outputs regularly, and every time the model changes.
  • These are our engineering practices, not a list of legal requirements for every business. For more on putting AI into production, see our LLM integration checklist and five lessons from our production RAG build.

    A short checklist before you ship an AI feature

  • Where is the entity that deploys it licensed: mainland, DIFC or ADGM?
  • Does the feature process personal data?
  • Are you in a regulated sector, such as banking, insurance or healthcare?
  • Have you written the user notice?
  • Which decisions need a human sign-off, and who gives it?
  • Can you produce a log of what the system did and why?
  • Who reviews the feature again when the model changes?
  • If you're in financial services, see how we approach fintech builds and our Trajex fintech case study. For health products, see healthcare. For an overview of our AI and Machine Learning work, see our services and software development in Dubai.

    Planning an AI feature?

    We help businesses build AI features with clear user notice, human oversight and careful data handling from day one. Book a free consult on WhatsApp +971 585692165.

    This article is general guidance, not legal advice. Confirm how these rules apply to your business with a qualified UAE lawyer.