Responsible AI: Governing the AI Already Inside Your Organization
Edition 01 · October 2026 · Coverage period: September 2026
Accendo Signals is a monthly guide to AI in emerging markets. It is written for impact investors, development banks, and the lenders, farm businesses and climate companies they support. Each issue covers what changed, where the money went, and what to ask next. This edition is about responsible AI: governing the AI already running inside your organization, so you know what it does and who is in charge of it.
Three Signals
- Kenya: Kenya’s central bank wants banks to guard against fraud and cyber attacks that use AI. Its draft rules are open for public comment until 7 November 2026.
- Money: Between 1 August and 25 September 2026, 91 percent of the AI money we tracked went into data centers and AI chips. One loan in Indonesia made up most of it, and without that loan the share drops to 51 percent.
- Supply: AI rented from a company can be switched off with little warning, and the best open-weight models, which anyone can download and run, now come from Chinese companies. Either way, knowing how fast you can switch to another model now matters.
Editor's Letter
Three Ways In
AI gets into a bank or lender in at least three ways, and most responsible AI policies only cover one of them. The first way is AI the lender chooses to build or buy. Examples are a tool that decides who gets a loan, a chatbot, or a system that sends repayment reminders. The second way is harder to see. AI now arrives inside normal software updates from the companies that run a lender’s banking system, credit checks and payments. The third way may be the biggest: staff use AI tools on their own phones and accounts because it helps them work faster.
I found it interesting that one of September’s biggest AI stories said little about any of this. Dario Amodei runs the AI company Anthropic. On 12 September he argued that frontier AI companies, the ones building the most powerful models, “must slow the pace at which we improve the capabilities of AI models”. Sam Altman of OpenAI and Elon Musk agreed within hours. For a lender serving people in Nairobi, Lagos, Chennai, Dhaka or Bogotá, most of the risk comes from the AI it is already using.
Each of the three ways can hurt the lender’s customers, the families and small businesses it serves. A loan tool can turn down a good borrower, and nobody can explain why. A software update can change the tone of repayment messages overnight. A staff member who pastes a customer’s ID card and mobile money records into a personal AI app has sent that customer’s private data outside the lender. For a lender, responsible AI means stopping these harms in all three ways, including in systems it never chose to buy.
The first steps are cheap. Ask five branch staff what they paste into AI tools. Write a one-page rule on which customer data must never go into a public AI tool. Ask every software supplier, in writing, which features use AI and which AI model sits behind each one. The board can then keep one list of all the AI in use, with a named person in charge of each item. Governing the AI already inside an organization starts with that list, and the lender stays responsible for it whatever its suppliers promise.
There is one more risk, and it is the supply of AI itself. Andrew Bailey chairs the Financial Stability Board (FSB), which brings together the world’s main financial regulators. He wrote to the finance ministers and central bank heads of the G20, a group of the world’s largest economies, in a letter published on 31 August. He wrote that “for the financial system, the most immediate concern is the potential impact of frontier AI on cyber risk”. He also warned about relying on a few big technology suppliers. AI you rent can be switched off, and the best open-weight AI comes mostly from one country (see the Explainer). So every lender needs a plan for losing an AI model.
Accendo Associates is writing a four-part series with a partner organization on how to make that plan. It starts with a simple design: one connection point between the lender and every AI model it uses. Through that point, the lender decides which customer data goes where and which model does which job.
The institutions best positioned for AI may not be those running the largest or newest models. They may be the ones that know what is already running inside them.
Prateek Shrivastava, Managing Partner, Accendo Associates
Portfolio Playbook
One practical plan an investor can run with the companies it backs this month.
Thirty Days to an AI List
This edition points to one first task for every company: find out what AI is already running, who is in charge of it, and what happens if it stops. An investor can ask for all three in one request. The plan below fits into a month for a small team. It uses the four supplier questions in Leaders and the switch table in the Explainer, both further down.
- Week one, the list. Write down every use of AI across the three ways in: what the company built or bought, what arrived inside supplier software, and what staff use on their own accounts.
- Week two, the suppliers. Send each supplier the four questions from Leaders, in writing, and add the answers to the list.
- Week three, the backup. Fill in the switch table for every AI service that affects a customer decision, and test one switch.
- Week four, the board. Show the board the list, a named person in charge of each item, and the first change the company will make.
The real test is the customer. A good result means fewer good borrowers turned down without a reason, customer data kept inside the company, and no gaps in service when repayments are due. Investors can ask for these four outputs at the next portfolio review.
Portfolio Playbook
Start with It Built In
Young companies can skip a step. A startup building its first products has no old systems to clean up, so it can set up its AI list, its supplier questions and its backup model in the first month, at almost no cost. An established lender doing the same work later has to untangle years of systems. That head start is real, and investors in early-stage companies, such as foundations and venture funds, can ask for it before they invest.
In August and September, eight early-stage AI companies raised USD 32 million between them, and each of them could start this way. Accendo works with investors and the companies they back, from early-stage ventures to established lenders, on this first step and on the AI plan that follows. If a company you invest in would benefit, write to prateek@accendoassociates.com.
Leaders
New responsible AI rules that can reach your portfolio companies through their suppliers and investors.
Two Rulebooks, One Vendor
Europe and the United States are handling AI rules in different ways. The European Union (EU) has one main AI law. In July 2026 it delayed its strictest rules, including the rules for AI that decides who gets credit, until 2 December 2027. One rule already applies, since 2 August 2026: people must be told when they are dealing with an AI system.
The United States has no national AI law. A December 2025 order from the President aims to set one national approach and challenges AI laws made by individual states. Some states still act on their own, and Colorado signed a new law in May 2026 on computer systems that make decisions about people. The Conference of State Bank Supervisors (CSBS) is the group for state bank regulators. On 16 September it released an AI checklist that bank inspectors can choose to use.
Lenders in emerging markets feel both sets of rules mainly through the software they buy and the investors who fund them. At the next supplier review, a lender can ask four questions in writing. Which features use AI? Which AI model sits behind each one? Where does customer data go, and is it used to train AI? How will the supplier warn us before the AI changes? Investors can ask their portfolio companies the same four questions once a year.
Leaders
Brazil's AI Law Waits
Brazil’s AI law is on hold. The lawmaker leading the bill (Bill 2338/2023), Deputy Aguinaldo Ribeiro, said in late August that Congress would vote on it only after the October elections. So Brazilian lenders will probably have no AI law to follow for the rest of 2026.
They still have one rule to follow today. Brazil’s data protection law (LGPD, Article 20) lets people ask for a review of decisions made only by a computer. Ask a Brazilian lender you invest in how it handles those requests for loan decisions right now.
Explainer
The Cost of Leaving
Most lenders do not make their own AI. They rent it from a company over the internet, or they use an open-weight model. Open-weight means the company publishes the model’s trained settings, called weights, so anyone can download it and run it on their own computers. Open-source goes further and also shares the data and code used to build the model. The cost of leaving is the time and money it takes to move a service from one AI model to another. That includes rewriting instructions, checking the new answers, and getting approval again if a regulator has already reviewed the old system. If the switch is slow, customers feel it, because loan applications get stuck or repayment reminders stop. That is why the cost of leaving is part of responsible AI, just like testing for unfair results or protecting data.
Rented AI can be switched off quickly. On 12 June 2026, Anthropic stopped access to two of its models, Fable 5 and Mythos 5, after an order from the US government under export rules. Fable 5 came back for everyone on 1 July, and Mythos 5 came back only for some US organizations. Open-weight AI cannot be switched off from outside, but it has its own problem, because the best options come from a small number of places. On 8 October, the top three open-weight models on the Artificial Analysis ranking all came from Chinese companies: Xiaomi, Z.ai and Moonshot AI. They scored 44 to 46 points, against 58 for the best rented model. The research group Epoch AI found in May 2026 that open-weight models are about four months behind the best rented ones.
One simple table solves most of this. It has one row for each AI service. Each row shows the model that runs it, the backup model, when the switch was last tested and how many hours it took. Investors can ask their portfolio companies for this table, and lenders can fill it in from their suppliers’ answers.
Advancements in Small AI
AI built to work with weak internet, unreliable power and older computers.
SEA-LION Runs Lighter
AI Singapore, a national program, and the chip maker NVIDIA released new AI models called Nemotron-SEA-LION-v4.8 on 18 September 2026. They are built for Southeast Asian languages. Each model is made of many small expert parts, and only a few of them work at any one time. That lets the smaller model run on modest computers: it has 30 billion parameters (the settings a model learns), but only about 3 billion are active at once. Both sizes also come in compressed versions that can run on devices close to the user. Since February, the same program has also offered SEA-Guard, safety models tuned to Southeast Asian languages and cultures.
Before backing a product built on these models, check the license, which the release does not state. Ask for test results in the customers’ own language. Also ask for the real cost per conversation on the device that will run it, measured outside a demo.
Capital Flows
Where AI money went, and which part of the AI system it funded.
One Loan, 82 Percent
Nuestros Capital Flows note of 28 September tracks AI deals in five regions of the world. Between 1 August and 25 September 2026, 25 signed deals added up to USD 3.79 billion. Of that, USD 3.45 billion went to computing power: data centers and AI chips. One deal was most of it. Zankore rents out AI computing power in Indonesia. It signed a loan of up to USD 3.1 billion with five banks to buy AI chips. That one loan is 82 percent of all the money. Without it, the total falls to USD 690 million, and the share going to computing power drops to 51 percent.
Companies that build AI products people actually use closed 15 of the 25 deals, with USD 169 million between them. The biggest was USD 125 million for Kapital Grupo Financiero in Mexico. Eight young startups raised USD 32 million between them, three grants added USD 11.5 million, and the rest did not say how much. The big loan is counted at its maximum, while the smaller deals are investments and grants. So this shows where the money went, and it is not a fair size comparison. Ten of the 25 deals did not share an amount, so all the percentages are based on the deals that did.
Southeast Asia: 3 signed deals worth USD 3.1 billion, almost all of it the Zankore loan.
Sub-Saharan Africa: 3 signed deals worth USD 470 million (see the Regional Spotlight).
Latin America and the Caribbean: 9 signed deals worth USD 206 million, the most deals of any region, led by Kapital Grupo Financiero.
South Asia: 4 signed deals worth USD 12 million, while large data center projects in Telangana and Odisha, in India, are announced but not yet signed.
Middle East and North Africa: 6 signed deals but only USD 1.5 million shown, because five deals by HUMAIN and White & Case did not share an amount.
Investors looking at a data center or AI chip deal can start with one question. How much of that computing power is booked by local companies? The Capital Flows Map is Accendo’s own record of publicly reported AI deals, and the note explains how we count them.
Regional Spotlight · Sub-Saharan Africa
Kenya Writes AI into Bank Rules
On 10 September 2026, the Central Bank of Kenya (CBK) asked the public for comments on new drafts of its main rule books for banks. Comments are due by 7 November. The drafts do not have a separate AI chapter. Instead, AI appears in three places. Boards should make sure AI-based early warning systems are in place. Internal auditors should check for fraud that uses AI. Security teams should defend against cyber attacks that use AI.
The CBK’s own survey, using data up to the end of 2024, found that half of the 125 institutions that replied were already using AI. Almost all of them, 93 percent, wanted the central bank to give guidance. That guidance has now arrived inside the normal bank rules, and customers will feel it most through fraud protection. Scammers can use AI to fake a voice or send convincing fake messages, and empty a customer’s mobile money wallet before the bank notices. Kenyan companies you invest in have until 7 November to comment. Two checks come first. Which of these rules apply to them, since microfinance banks and digital lenders have separate rules? Does this year’s audit plan already cover fraud that uses AI?
Most of the region’s signed money went to infrastructure. Three deals were tracked in August and September, worth USD 470 million. The biggest was a USD 300 million investment in the West Indian Ocean Cable Company (WIOCC), a digital infrastructure company. It came from the Africa Finance Corporation and Vision Invest. The second was a USD 170 million loan from Korea’s Economic Development Cooperation Fund. It will build an AI and digital skills institute in Tanzania, one of the few signed deals for skills. Since January 2023, the map has found only 30 deals in Sub-Saharan Africa for anything beyond infrastructure. They span 23 countries and add up to about USD 110 million (Necesario, pero no suficiente). When a company you invest in plans an AI product, ask how much of the budget goes to local data and staff training. Then ask who is funding that part.
Southeast Asia: The Zankore loan makes up almost all of the region’s signed money. The SEA-LION release (see Advancements in Small AI) brought new AI models for the region’s languages.
South Asia: Reuters reported on 1 September, citing unnamed sources, that the National Payments Corporation of India (NPCI) is preparing a new system. It would let AI assistants make small payments on UPI, India’s instant payment network, without the user approving each one. Pine Labs, an Indian payments company, has run a similar system since June, which needs the customer’s permission only once.
Latin America and the Caribbean: CAF, the development bank for Latin America and the Caribbean, has announced a USD 12 billion plan for digital change. It is not yet signed.
Middle East and North Africa: Egypt’s plan for a national AI data center is also announced and not yet signed.
Overheard
“An institution may outsource the computation, but it cannot outsource the consequence.”
Rohit Jain, Deputy Governor, Reserve Bank of India, keynote at Global Fintech Fest, Mumbai, 9 September 2026.
What We're Watching
Three deadlines. The FSB’s final report on twelve good practices for responsible AI is due in October 2026. Colorado’s comment period on its new AI rules ends on 26 October, and Kenya’s comment period ends on 7 November.
One panel. On 28 October at 08:00 New York time, Accendo hosts a panel on getting more investment into African AI beyond data centers (register here). The speakers are Nick Williams (African Development Bank), Johan Bosini (Quona Capital) and Snehar Shah (iXAfrica). The panel picks up where the Spotlight leaves off.
Two tests. We will watch whether the eight biggest unsigned AI plans on the Capital Flows Map, worth USD 66.6 billion, turn into signed deals. We will also watch whether NPCI confirms its system for AI payments.
