GAAN WEEKLY AI INTELLIGENCE BRIEFING – 31 October 2026
GAAN WEEKLY AI INTELLIGENCE BRIEFING – 31 October 2026
For Members of the Global AI Advocates Network (GAAN) Government Officials, NGO and Business Leaders & Senior Policy Influencers
Week ending Saturday 31 October 2026 | Published by GAAN | www.global-ai.solutions
Webinar and Podcast Update
As indicated in a previous newsletter, we have discontinued our twice-monthly webinars. Articles on the most important and urgent issues are available on Alan Ross's LinkedIn profile, and videos are available on our YouTube channel @AIGrandad999.alanross and on our 10 language channels (e.g. Spanish, French, Hindi, Swahili, Arabic, Bangla, Chinese, Russian, Serbian, Indonesian) where there are now over 25 long and 12 short videos on each channel covering the AI and technology disruptions that will affect your country. We encourage you to view and share these with your networks and to ask them to subscribe so they are notified whenever a new video is uploaded. We have now signed a memorandum of understanding (MoU) with the prestigious Centre for Artificial Intelligence and Sustainable Development (CAISD), based in South Africa, which applies artificial intelligence to sustainable development in Africa. Its work spans agriculture, mining, governance, environment and healthcare, and it works with a consortium of African partner universities and international partners. We have agreed to cooperate so we can extend our activities across the 54 African nations while continuing to support and inform other regions across the Global South. Check out the link below, which shows the recent AFRICAST video podcasts that we have done through CAISD. CAISD podcasts and videos: https://www.youtube.com/channel/UC9m_MM9O0Mkzw0cKXcswaxw The audio podcast outlines how countries can fund AI adoption, and the in-depth video interview podcast is wide-ranging, covering important issues of relevance to all countries. Discussions are ongoing, but it is likely that we will do one podcast each month focused on the 54 African countries and one each month for a global audience covering Asia, Africa and Latin America, as well as audio podcasts on various topics. The topics discussed, whether in the African podcast or the global podcast, will be relevant for all countries in the Global South. This Week's Key Developments A regional sweep of the week from Sunday 25 October to Saturday 31 October 2026, with developments from Africa, Asia and Latin America, plus two global stories with strong Global South relevance. Key points only below; follow the source links for the full story. AFRICA 1. TALENT NATION AIMS TO TRAIN 1 MILLION AFRICAN AI ENGINEERS Lyinoluwa (Iyin) Aboyeji, co-founder of Andela and Flutterwave, has launched Talent Nation, a startup that trains African AI engineers and places them with companies around the world, with the aim of training one million engineers over the next ten years (reported 28 and 29 October). It has raised an undisclosed preseed round co-led by Enza Capital and London-based LocalGlobe, with Future Africa, Kaleo Ventures, Ajim Capital, Solving.org and Accelerate Africa also taking part. IHS Towers is named as a client. Trainees pay nothing upfront. Client companies pay for the engineers they hire, that revenue funds new trainees, and employed engineers later help pay for future cohorts. Candidates take an online test and then in-person trials; the strongest are placed with clients immediately, while others join 3-, 6- or 12-month programmes that combine projects, internships and work experience. Its predecessor, Learn2Earn, has received over 63,000 applications, assessed over 32,000 candidates and admitted about 1,400 fellows (under 3% of applicants) across nine campuses in six cities. In February, Learn2Earn partnered with the Digital Bridge Institute to train AI engineers in support of Nigeria's 3 Million Technical Talent (3MTT) programme. No detailed timetable for the one million target has been published. Why this matters for GAAN members: A "train now, employers pay later" model shows how large AI skills programmes can be financed without big upfront public budgets. Governments can partner with programmes of this kind to meet national targets such as Nigeria's 3MTT, and link them to National AI Entrepreneurship Training Initiatives (NAETIs) in every country as proposed by Alan Ross. Sources: iAfrica, 29 Oct 2026; Startup Researcher, 28 Oct 2026 2. SOUTH AFRICA'S REDRAFTED NATIONAL AI POLICY NOT EXPECTED BEFORE MARCH 2027 Jeanette Morwane, acting deputy director-general at the Department of Communications and Digital Technologies and the policy's project manager, told a South Africa–Finland AI dialogue in Pretoria that Cabinet approval to publish the revised policy for public comment is now expected no earlier than March 2027, months later than first promised (reported 31 October). The first draft was gazetted on 10 April 2026 and withdrawn on 26 April because its reference list contained fictitious sources. The drafting team must submit the revised policy and a discussion document to the minister by 30 November, followed by internal government consultations from December to February. A seven-member independent expert panel chaired by Wits University AI researcher Benjamin Rosman has submitted advisory notes, and the government drafting team includes the Presidency, the Department of Science, Technology and Innovation, the State Security Agency and the Department of Trade, Industry and Competition. An implementation road map is being drafted in parallel. Discussions centre on compute and data sovereignty and on "model sovereignty", meaning large language models that handle South Africa's official languages. Ms Morwane said significant investment in computing infrastructure is needed. Why this matters for GAAN members: A clear warning for every ministry: references and evidence in AI-assisted drafts must be checked, because unverified sources can set a national policy back by a year. Officials can also learn from South Africa's combination of an independent expert panel, a cross-government drafting team and a parallel implementation road map. Source: iAfrica, 31 Oct 2026 3. LESOTHO DEVELOPING AN AI CHATBOT TO ADVISE FARMERS Lesotho's Minister of Information, Communications, Science, Technology and Innovation, Nthati Moorosi, announced at the opening of the third National Science, Technology and Innovation Week in Maseru that the government is building an AI chatbot to give farmers easier access to agricultural information and advice (reported 26 October). It is being developed with the Ministry of Agriculture, Food Security and Nutrition. The government is also upgrading high-performance computing at the Mohale's Hoek Data Centre, and a Research and Innovation Policy is awaiting approval. No launch date, languages or number of farmers to be served have yet been announced. Why this matters for GAAN members: A small, low-income country is starting with one practical, high-impact public service. Agriculture ministries elsewhere can follow this approach, pairing a farmer advisory chatbot with national data and computing capacity, and planning from the start for local languages and basic phones . Source: iAfrica, 26 Oct 2026 4. HONEYWELL LAUNCHES INDUSTRIAL AI AT NIGERIA'S DANGOTE REFINERY Honeywell has launched Forge Plant Cognition, a new industrial AI platform built on NVIDIA and Dell Technologies capabilities, at the Dangote refinery in Lekki, Nigeria (reported 31 October). The platform is designed to improve productivity and profitability in complex industrial plants. Based on preliminary data, Honeywell estimates benefits of more than 29 US cents per barrel, or roughly US$60 million a year in gross operational benefits at full-capacity throughput. The estimate is not guaranteed and excludes implementation costs. The refinery's nameplate capacity is 650,000 barrels per day, and Honeywell is helping Dangote expand it to 1.4 million barrels per day by 2028. Why this matters for GAAN members: Large industrial AI systems are now arriving in African plants. Governments can use major projects like this to build local skills, by requiring training and knowledge transfer as a prerequisite in any contract. Look look for similar efficiency gains in state-owned utilities and plants. Source: iAfrica, 31 Oct 2026 ASIA 1. INDIA'S 10,000-GPU TENDER FOR SHARED AI COMPUTE CLOSES India's Centre for Development of Advanced Computing (C-DAC), under the Ministry of Electronics and IT, closed submissions on 29 October and was scheduled to open technical bids on 30 October in its IndiaAI Mission tender to empanel providers of up to 10,000 AI compute units (GPUs, TPUs and other accelerators) over five years. It follows Minister Ashwini Vaishnaw's announcement on 11 October that 10,000 more GPUs would be added to the mission's shared compute pool. Clusters of 4,000, 1,000 or 512 units must each be housed in a single data centre in India, for model training, fine-tuning, inference and other AI workloads. Intended users include ministries and departments, academia, research institutions, startups and industry. Reports of the minister's announcement say 3,000 units are reserved for government and public-interest work and 7,000 offered to companies. More than 38,000 GPUs have already been added to the IndiaAI compute platform through earlier rounds and offered at subsidised rates, although startups and enterprises still report waits of several months for high-end capacity. Why this matters for GAAN members: India contracts private providers and then offers subsidised shared access, instead of building one national supercomputer. Officials elsewhere can study the tender design, including minimum cluster sizes, local hosting and a set-aside for public-interest work. Perhaps even more importantly shortage of compute capacity is making Global companies overseas come to Indian data centres. Is this a new role for the developing world ? Sources: Elets CIO, 30 Oct 2026; MediaNama, 13 Oct 2026 2. INDIA'S BIHAR STATE APPROVES ITS ARTIFICIAL INTELLIGENCE POLICY 2026 The Cabinet of Bihar approved the Bihar Artificial Intelligence Policy 2026 on Wednesday 28 October to promote AI-based governance, innovation and digital enterprises. Additional Chief Secretary Arvind Chaudhary said it offers financial incentives for AI companies, global capability centres, startups and digital enterprises, including capital investment and interest subsidies, support for cloud computing, lease rental and electricity charge support, employment generation incentives and SGST reimbursement. The policy also promotes AI applications in agriculture, healthcare, education, manufacturing and urban services. At a review of the draft policy on 22 October, the state government set a target of AI training for 500,000 (5 lakh) students over five years, with a major AI event each year for the next five years. Why this matters for GAAN members: Bihar shows that state and provincial governments can act on AI, not only national ministries. Sub-national officials elsewhere can adapt this package of business incentives combined with a large student training target – very much in line and supportive of our efforts to promote the need for a Global AI entrepreneurship Training Initiative ( GAETI) and national programmes ( NAETIs) as we have been advocating through our World AI Academies Association ( WAAA) Sources: UNI India, 28 Oct 2026; NewKerala, 22 Oct 2026 LATIN AMERICA AND THE CARIBBEAN 1. MICROSOFT BACKS LATAM GPT, THE REGION'S OPEN LANGUAGE MODEL Microsoft and Chile's National Centre for Artificial Intelligence (CENIA) have signed a two-year memorandum of understanding to support Latam GPT, the first large open language model created in and for Latin America and the Caribbean (announced 28 to 29 October). It was signed by Juan Lavista Ferres, Microsoft corporate vice president, chief data scientist and director of its AI for Good Lab, and Álvaro Soto, director of CENIA. In the first phase Microsoft will provide US$250,000 in Azure cloud credits, usable until February 2027, for Latam GPT's computing infrastructure and operations, together with joint research and technical support from its AI for Good Lab. The partners will also develop joint case studies in education, public services and health. Latam GPT, presented in February 2026, was built with more than 300,000 million tokens of Spanish and Portuguese text, with around 60 institutions from 15 countries taking part. Its promoters say Spanish and Portuguese make up only about 4% and 2% of the data used to train large AI models. The agreement supports development; it is not yet a finished application for businesses or users. Why this matters for GAAN members: Pooling effort across 15 countries has produced a regional open model at modest cost. Governments in Africa and Asia can consider similar regional models for shared languages, and Latin American and Caribbean governments not yet involved can join and contribute public data. Sources: CENIA, 29 Oct 2026; TeleSemana, 30 Oct 2026; CNN Chile 2. MEXICO TECH WEEK 2026 PUTS AI AT THE CENTRE The fourth Mexico Tech Week is running from 26 October to 1 November 2026 across Mexico City, with more than 500 events and about 30,000 expected attendees. Tecnológico de Monterrey is the strategic partner. AI activities include an AI House dedicated to AI and business applications, and the AI for Corporates Challenge, a competition in which AI startups present solutions to real company problems. On 27 October the AI House brought together investors from SoftBank, Kaszek and General Atlantic to discuss AI investment and funding trends in Latin America. Organisers say they want to show that Mexico can create AI, not only consume it. Why this matters for GAAN members: Large, open ecosystem weeks connect startups, investors and companies at little public cost. Ministries and city governments can support similar events to link local AI startups with buyers and investors to strengthen and their eco systems for AI entrepreneurs and startups . Sources: ADN40, 13 Oct 2026; DPL News GLOBAL 1. LIQUID AI RELEASES SMALL OPEN MODELS THAT RUN ON PHONES AND LAPTOPS On 27 October, Liquid AI released LFM3, a family of six open-weight models of about 2 billion parameters covering text, vision, audio, combined ("omni") input and decision tasks, all built on one shared base model. The models are available on Hugging Face, the Liquid AI Playground and the Liquid AI API. The main LFM3-1.7B model has a 256,000-token context window and needs about 1 GB of memory at 4-bit precision, so it can run on ordinary laptops and phones. The LFM Open License v1.0 allows free commercial use for organisations with less than US$10 million in annual revenue; larger organisations need a separate agreement. Performance figures so far are vendor-reported, and independent evaluations had not been published at release. Why this matters for GAAN members: Small models that run without the cloud make "small AI" services in health, farming and local government possible where connectivity and budgets are limited. Officials should include on-device AI in digital strategies and procurement plans. Sources: Unite.AI, 27 Oct 2026; MarkTechPost, 27 Oct 2026; i-SCOOP 2. TWO IN THREE LEARNERS USE AI CHATBOTS TO PREPARE FOR EXAMS, OXFORD SURVEY FINDS A new Oxford University Press report, Digital Learning: Insights from the Classroom, covering South Africa, Brazil, the UK and Hong Kong, found that 66% of learners use AI chatbots to prepare for exams (reported 29 October). Most learners see chatbots as fair for revision, planning and generating ideas, but not for doing the work itself. Among teachers, 82% use AI chatbots for research, 74% to create resources, 66% for administration and 56% for marking; only 3% never use them. The device gap is widest in South Africa, where 18% of learners reported having no digital device in at least one setting, compared with 13% in the UK and 4% in Brazil. The report summary did not give the sample size. Why this matters for GAAN members: AI is already in classrooms in developing countries, with or without policy. Education ministries need clear guidance on acceptable AI use in assessment and should close device gaps so that learners in poorer regions are not left behind . Source: iAfrica, 29 Oct 2026 Not yet part of GAAN? The not for profit Global AI Advocates Network (GAAN) connects senior government officials, Academics ,NGO and business leaders across the developing world who are working to ensure AI eliminates rather than widens the gap between nations. Join our growing network across the developing world, to expand our 50+ connected countries towards our target of covering 135 low- and middle-income countries( LMICs) We need officials from more governments to join us so they can be kept informed of developments where there may be opportunities to share knowledge or expertise across borders . Please share this newsletter with one or two relevant government officials in your country and especially in other countries so we can increase our network to keep all countries informed of important developments of relevance Contact alan.ross@global-ai.solutions View and share our videos on important issues in 11 languages on our YouTube channels (as indicated in the opening paragraphs) and listen and view our latest podcasts with CAISD https://www.youtube.com/channel/UC9m_MM9O0Mkzw0cKXcswaxw #AIforDevelopment #AIGovernance #GAAN #DigitalDivide #GlobalSouth #AIforGood #LMICs #AIPolicy #SmallAI #AISkills #CAISD #Africa #Asia #LatinAmerica