GCC AI News — 18 Oct 2026

On 15 October, MBZUAI's Institute of Foundation Models and Inception42 launched Jais 3, a fleet of Arabic-native AI models. It pairs six open language models, from 900 million to 375 billion parameters, with Jais Speech, a recognition model covering Modern Standard Arabic and dialects spoken across 19 Arab countries. The models understand spoken and written Arabic across dialects, follow speakers who switch between Arabic and English or French mid-sentence, and carry out practical work from transcription to building documents and presentations. Jais Speech recorded the lowest average word and character error rates of any model compared, and all models are available today on Hugging Face with a 512,000-token context window, the longest of any Arabic-native model.

Generative AI has long underserved the more than 400 million Arabic speakers. An open, dialect-aware fleet built in the UAE gives governments and businesses sovereign models they can adapt to their own hardware and budget, reflecting the country's long-term investment in Arabic AI research.

An Emirati government call centre can deploy Jais Speech to transcribe citizen calls in Gulf dialect in real time, then use a Jais 3 language model to draft a reply or summary, keeping the whole workflow in Arabic and on local infrastructure.

On 12 October, HUMAIN, the Public Investment Fund's full-stack AI company, announced a strategic partnership with frontier AI firm Napster to bring advanced Arabic-native AI experiences to public and private organisations across the Middle East. The deal combines HUMAIN IQ, HUMAIN's AI services platform, and HUMAIN Voice with Napster's video presence, Large Persona Model technology and multimodal agent infrastructure, letting the companies deploy Arabic-fluent digital employees at scale. HUMAIN CEO Tareq Amin framed it as an example of Saudi-developed technology enhancing an international company's capabilities, with technology companies able to integrate HUMAIN's models through secure APIs.

The Gulf's AI story is usually about importing compute and models. This flips it: Saudi-built Arabic AI becomes a product other companies license and build on, a sign the Kingdom is moving from AI consumer to AI exporter under Vision 2030.

A regional bank or retailer can plug HUMAIN's Arabic language and voice models into Napster's agent platform and roll out culturally fluent digital customer-service agents, without building Arabic AI capability from scratch.

On 16 October, Abu Dhabi's Department of Government Enablement announced the establishment of TAMM Global, a company created to bring the emirate's AI-powered unified government services platform, TAMM, to other governments. Under its founding law, TAMM Global will develop, manage and deliver an integrated ecosystem of digital services, own and operate the underlying platforms and infrastructure, and hold the associated intellectual property. DGE Chairman Ahmed Tamim Hisham Al Kuttab said TAMM was built in Abu Dhabi to improve quality of life, now serves millions of users, and that the new company makes capabilities proven in Abu Dhabi available to governments seeking better outcomes for their people.

Most countries buy digital-government technology from foreign vendors. A Gulf government turning its own proven, AI-driven services platform into an export shows home-grown public-sector innovation reaching global markets, while advancing Abu Dhabi's goal of becoming the world's first AI-native government.

Another government can adopt TAMM's platform to give its citizens a single AI-powered app for licences, payments and permits, instead of spending years and large budgets building a digital-services stack from the ground up.

A team led by MBZUAI researchers won first prize at the Robotic Urology Breakthrough Innovation Challenge at ERUS26, Europe's largest robotic-urology meeting, for a system called Janus AI. It is a video-to-video retrieval tool: it matches footage captured live in the operating room against a library of 30-second clips from similar surgeries and puts the reference material in front of the surgeon as the operation unfolds. Features for every library clip are precomputed, so only the live clip is processed in the moment and the similarity search runs in RAM in milliseconds. The work, reported on 13 October, is a collaboration between MBZUAI, the University of Illinois Chicago and the University of Turin, with surgeons able to keep watching full-length reference videos.

Surgical skill is hard to transfer. A tool that surfaces how the best surgeons handled a comparable moment, instantly and during the procedure, is a concrete educational aid for residents and less experienced surgeons, and a Gulf research team's work recognised on a global stage.

A junior urologist facing an unusual situation mid-operation can see 30-second clips of how expert surgeons handled the same step, limited to the best recorded procedures, helping them decide their next move with more confidence.

MBZUAI announced DoAtlas-2 on 7 October, a causal world model of human biology that runs the full research cycle on its own, from forming a question to testing a hypothesis against real human data. It draws on 771 research resources covering more than 720,000 participants across 48 countries, spanning clinical, physiological, imaging and molecular data, and holds a causal model of over 93,000 biomedical concepts and nearly 150,000 candidate relationships. It has evaluated 2,031 research questions end to end. Crucially, study designs are frozen before results are revealed, failed tests are kept to steer analysis, and every claim traces back to its evidence. MBZUAI plans an open-source release starting with 1,000 research questions and 30,000 results.

AI can now generate hypotheses faster than scientists can test them. A system that validates which ones hold up in real data, with transparent, traceable reasoning rather than black-box outputs, targets a genuine bottleneck in medicine, work led from a Gulf university.

A medical research group can use DoAtlas-2 to screen thousands of candidate links between, say, blood markers and vascular outcomes, and prioritise only the hypotheses that survive testing in real population data for costly lab follow-up.

The Qatar Investment Authority announced it has joined Positron AI's $875 million Series C financing, which values the AI inference-hardware company at $5 billion. The round was co-led by NEA, Atreides Management, Valor Equity Partners, Andra Capital and others, and QIA had also taken part in Positron's Series B in February 2026. Positron builds hardware and software to make AI inference, the process of running trained models, more cost- and energy-efficient, using a memory-first architecture that reduces reliance on constrained high-bandwidth-memory supply chains. The new funding supports Positron's next-generation silicon, Asimov, and the production ramp of its multi-terabyte-memory inference system, Titan.

Inference is becoming the dominant cost of running AI as agents and assistants proliferate. A Gulf sovereign fund backing a company focused on cheaper, more efficient inference ties the region's capital to the infrastructure layer that will shape how affordable AI becomes worldwide.

Data-centre operators, including those in the Gulf, could eventually deploy Positron's memory-first systems to run large models at lower power and cost, easing the hardware constraints that currently limit how much AI they can serve.

Bahrain's Labour Fund, Tamkeen, has launched an enhanced Digital Enablement and AI Program that pays up to 50 percent of the cost of software and digital tools for micro, small and medium-sized enterprises. The programme is designed to help businesses across sectors bring AI and other digital technologies into day-to-day operations, covering areas such as automated inventory management, predictive accounting, computer-vision quality control and bilingual customer-service bots. Tamkeen lets firms choose and implement tools based on their actual operational needs rather than a one-size-fits-all package, and the initiative complements its broader national goal of training 50,000 Bahrainis in AI by 2030, aligned with its 2026 strategic priorities.

Cost is the biggest barrier stopping small firms from adopting AI. Co-funding half the bill directly addresses that, helping prevent a productivity gap between well-capitalised multinationals and local merchants, and spreading the benefits of AI across Bahrain's wider economy.

A Bahraini retailer can adopt an AI-powered inventory or customer-service tool and claim up to half the cost back from Tamkeen, making an upgrade it would otherwise have delayed for a year financially viable now.

Oman has established an Artificial Intelligence Special Zone in the Governorate of Muscat, created by Royal Decree 50/2026 and overseen by the Ministry of Transport, Communications and Information Technology. The zone is designed as an integrated, innovation-driven ecosystem that attracts and nurtures promising technology startups by providing advanced, subsidised tech infrastructure, tax exemptions and a supportive regulatory environment. It is a civilian economic initiative aimed at building a home-grown AI and digital economy in the Sultanate, offering licensed entities incentives to set up and scale, and forms part of Oman's broader push under Oman Vision 2040 to diversify beyond hydrocarbons into a knowledge-based economy.

Dedicated zones with real incentives are how governments turn AI ambition into companies and jobs. By lowering the cost and regulatory friction of starting an AI business, Oman gives founders a concrete reason to build in the Sultanate rather than abroad.

An AI startup can set up inside the Muscat zone to access subsidised computing infrastructure and tax exemptions, lowering its running costs in the critical early years while it builds and commercialises its product.

Researchers at MBZUAI in Abu Dhabi presented work, reported on 14 October, showing how genetic risk and biological responses can differ between populations, and why the genomes missing from today's datasets matter for everyone's health. Much of the genetic data underpinning modern medicine comes from a narrow set of populations, which can leave diagnostic tools and treatments less accurate for people whose ancestry is underrepresented. The research, part of MBZUAI's growing genomics and health effort including a long-term deep-phenotyping study in Abu Dhabi, argues that broadening the genetic record using AI to analyse more diverse data could improve predictions of disease risk and drug response across populations that medicine has historically served less well.

AI in healthcare is only as fair as the data behind it. A Gulf university pushing to include underrepresented populations in genetic research tackles a real equity gap, and could make diagnostics and treatments work better for people across the region and beyond.

A clinician treating a patient from an underrepresented population could, over time, rely on risk models and drug-response predictions tuned on more diverse genetic data, rather than tools calibrated mainly on a narrow group.

On 13 October, Dell Technologies and UAE firm Tahaluf Al Emarat Technical Solutions signed a Memorandum of Understanding to deploy AI-powered solutions that help the public sector improve urban services and respond to residents' and visitors' needs. The collaboration combines Dell's AI and edge-computing capabilities with Tahaluf's in-vehicle technology platform, putting processing close to where data is captured. That lets city organisations gain faster visibility into issues across large areas and act sooner, supporting quicker identification of road and infrastructure problems and parking violations while reducing reliance on constant cloud connectivity. Dell's Walid Yehia said AI is helping cities become more responsive, and Tahaluf CEO Dr Ahmed Musabbeh Al Kaabi called it a step toward smarter, more efficient urban experiences.

Smart-city promises often stall on connectivity and latency. Processing data in the vehicle, at the edge, lets services react in near real time across a sprawling city, a practical application of AI that residents feel directly through better-maintained roads and services.

A municipal fleet fitted with edge AI can automatically flag a pothole or a blocked lane as it drives past, so the city dispatches a repair crew sooner instead of waiting for a resident complaint.