Fifteen years of living and working in the Gulf teaches you things that no briefing document can. One of the first is the rhythm of the year. Not the Gregorian calendar that most of my clients in London or New York still default to, but the real rhythm: the way Ramadan changes everything, the quiet of a Friday morning in a city that on any other day feels like it never sleeps, and the particular warmth of Eid week, when business cards get put away and what matters is who you are, not what you do.
I carry that rhythm with me now. It has become part of how I read situations, how I notice when something is off. And when I started paying close attention to how AI tools were being deployed in marketing and communications across the region, that instinct started firing constantly.
The tools were foreign. Not in their interface, but in their assumptions.
Large language models and the image generators now embedded into most marketing workflows are trained predominantly on Western, English language data. The academic literature on this is well established. But the lived experience of it in the GCC is more commercially specific, and more consequential, than the research tends to capture.
Ask most AI marketing platforms to generate content for Eid al-Adha and what you receive is approximation. The visual language is borrowed from adjacent references. The emotional register is slightly off, close enough to not be obviously wrong, but not close enough to feel true. The greeting is technically correct. The understanding behind it is not there. It is the difference between someone who has learned to say Eid Mubarak and someone who knows what it means to say it, and to whom, and in what tone, and with what weight of relationship behind the words.
This is not a marginal problem. The global Muslim consumer market represents an estimated $2.8 trillion in purchasing power. The GCC alone concentrates some of the world’s highest per capita spending. Brands that get the cultural moment right build something that money cannot easily replicate: trust, belonging, the felt sense that they see you. Brands that get it wrong during Eid week do not simply miss an opportunity. They communicate something about who they were built to understand, and who they were not.
The governance and legal frameworks are beginning to catch up, slowly. The CBUAE’s 2026 AI guidance requires annual bias testing for AI systems in licensed financial institutions. The EU AI Act contains provisions on non discrimination and fundamental rights. But neither framework is yet asking the harder cultural question, which is not whether AI systems discriminate in the narrow legal sense, but whether they exclude and flatten in the broader human one. Whether a system trained to understand one world can serve another with genuine competence, or only with the appearance of it.
The region itself is not waiting. From Abu Dhabi, Jais 2, developed by Inception and the Mohamed Bin Zayed University of Artificial Intelligence, was built specifically to embed Arabic cultural and dialectal fluency at its foundation, trained on the richest Arabic first dataset released to date. Saudi Arabia’s HUMAIN has launched ALLaM 34B, fine tuned by over six hundred domain experts and designed around explicit Islamic values and cultural alignment. Qatar’s Fanar 2.0, developed at the Qatar Computing Research Institute, follows the same logic. And PalmX 2025 introduced the first benchmarking framework designed specifically to assess large language models on Arabic and Islamic cultural competence across twenty two Arab countries. These are not research side projects. They are sovereign infrastructure investments. The gap the Georgetown data identifies is being treated, in the GCC at least, as a strategic problem worth solving at the model level, not patching over at the prompt level. For brands operating in the region, the question is no longer whether culturally fluent AI is coming. It is whether they will build for it, or wait until it is already the baseline.
I think about this from a specific position. Fifteen years of working between legal systems, between languages, between expectations that do not always translate, has made one thing very clear to me: perspective is never neutral. Every framework, every process, every algorithm, carries the assumptions of whoever built it. The question is not whether your AI has a point of view. It does. The question is whose point of view it is carrying, and whether anyone in your organisation has interrogated that before putting it to work.
The brands getting this right in the GCC are not the ones spending the most on Eid campaigns. They are the ones who have built enough genuine cultural intelligence into their teams that the AI remains a tool in informed human hands, rather than the decision maker itself. They know what the algorithm cannot know. They fill the gap.
The algorithm does not celebrate Eid. That is not a criticism. It is a design fact. The responsibility, legal, ethical, and commercial, lies with the people who deploy it, and whether they have been honest with themselves about what it understands and what it does not.
THE SIGNAL
The global Muslim consumer market is estimated at $2.8 trillion in annual purchasing power. Yet fewer than 12% of major AI training datasets contain substantive representation of Islamic cultural content, according to research from the AI and Faith initiative at Georgetown University. The gap between market significance and model comprehension is not a technical footnote. It is a strategic and governance failure hiding in plain sight.
THE QUESTION
If your AI tools were audited today, not for legal compliance, but for cultural competence, what would they reveal about whose world you built them to understand?
Sources & Further Reading
AI and Faith Initiative, Georgetown University: AI Training Data and Religious Representation (2025)
CBUAE AI/ML Guidance for Licensed Financial Institutions (February 2026)
EU Artificial Intelligence Act: Non-Discrimination and Fundamental Rights Provisions
DinarStandard: State of the Global Islamic Economy Report 2025/2026
Mindfoundry: AI Regulations Around the World (2026)
- Inception, Cerebras and MBZUAI: Jais 2 Release (December 2025)
- HUMAIN / Saudi Data and AI Authority: ALLaM 34B Launch (2025)
- Qatar Computing Research Institute: Fanar 2.0
- PalmX 2025: Benchmarking LLMs on Arabic and Islamic Culture (arXiv)
Running a business is hard. Building a website shouldn't be.
With Readdy.ai, just describe your business, the AI builds your website.
No coding or design skills needed. Launch in 2 minutes.