SDAIA and SASO take Saudi AI toward global standards

Global standards are the prize Saudi Arabia is now chasing in artificial intelligence, not only another ranking. The Kingdom marked 2026 as a Year of Artificial Intelligence. In mid-September it co-hosted UNESCO’s fourth Global Forum on the Ethics of AI in Riyadh, the first edition in the Gulf, with the Saudi Data and AI Authority and the International Centre for AI Research and Ethics. A ministerial statement on 17 September called on states to move UNESCO’s ethics recommendation from principle into national law, capacity programmes and cross-border cooperation. That is committee work. It is how countries try to write the rules other people later have to follow. That is the punch. Hosting a forum is not the same as owning ISO. The closer institutional moves matter more. The Saudi Standards, Metrology and Quality Organization took BSI certification for ISO/IEC 42001, the first international AI management-system standard, and was the first Saudi body to do so in March 2026. The environment ministry later posted the same certificate. SASO also won a seat on ISO’s Technical Management Board for 2026 to 2028. That board shapes how international standards get scoped. A seat is not a veto. It is a microphone in Geneva that Riyadh did not have. Law is moving in parallel. Implementing regulations to the updated Copyright Law, published on 31 July 2026 and in force from mid-August, set out how copyrighted works may be used to develop AI products and algorithms, and added a compulsory-licence path for Arabic translations when rights holders stay silent. Law firms already flag gaps on generated output and on how wide the training exception runs. The intent is clear. Arabic data and Arabic models are a standards fight as much as a product fight. If the rulebook for training on Arabic text is written in Riyadh, labs that want that corpus have to read it. SDAIA remains the centre of gravity. NSDAI still frames the national bid: talent, regulation, open data, startups, investment. SAMAI, the million-Saudis training drive, has passed its headcount target, with ethics and safe-use modules inside the curriculum, Ahmed Al-Ghamdi of SDAIA’s capacity-building arm told Arab News at the UNESCO forum. Rankings are the poster the government prefers. A UN-linked global AI index placed the Kingdom 14th of 83 countries and first in the Arab world, still first on government strategy, and seventh on AI trade, after a 17-place jump. IMD’s 2026 yearbook had it 13th of 70 and third among G20 peers, with high marks for company adoption and public access. Strategy-first scores are easier to ace when you built SDAIA by royal order. Research and model-frontier scores are harder. The MENA angle is language and venue, not another compute park. Gulf states buy GPUs. Few of them sit on ISO’s technical board or write a copyright exception aimed at Arabic large models. A Riyadh ethics forum plus ICAIRE plus 42001 certificates is an attempt to make Gulf practice look like the template for states that missed the EU AI Act and will not copy California. Egypt, the UAE and Qatar will either plug into that template or publish their own. Fragmented Gulf rulebooks would cancel the point. Caveats should sit next to the Year of AI branding. Global standards for foundation models are still drafted in labs and in Brussels, Washington and Beijing. ISO 42001 governs how an organisation manages AI. It does not decide model evals. A ministerial communiqué does not bind OpenAI or DeepSeek. Training a million people on ethics slides is not a red-team. Copyright text that still needs SAIP guidance will be tested in court, not at a summit. If Saudi drafts that other regulators cite, and if Arabic training rules become the default for MENA models, the standards bid is real. If the output stays rankings, badges and closing statements, this was a loud year of positioning. Count adopted clauses, not stages.