On this page
TL;DR:
- AI training for East African companies is best planned as one shared core (safe habits, good prompts, checking every output) with a short local layer for each country's data law, language and connection.
- Data law differs sharply. Tanzania requires a permit before personal data leaves the country. Rwanda's default is storage in Rwanda. Uganda needs no approval per transfer but expects records. South Sudan had no dedicated data protection law in the latest sources we found.
- As of 8 November 2026 we found no dedicated AI law in force in any of the seven countries covered here. Data protection law already applies to AI use wherever it exists.
- English is the safest working language for AI. A fluent colleague should own every output in Kiswahili, Amharic, Somali, Kinyarwanda or any other language.
- Set one written policy using the strictest rule for shared work, and treat the training itself as a data flow: participant lists cross borders too.
Who this guide is for
This guide is for HR, learning and operations leaders at companies with staff in more than one East African country. It covers Kenya, Uganda, Tanzania, Rwanda, Ethiopia, Somalia and South Sudan.
You do not need a technical background. If someone has asked you to "roll out AI training across the region", this should help you scope it, budget your time and question providers.
For one country in depth, we have separate guides for South Sudan, Tanzania, Uganda and Somalia.
What regional teams actually need
Regional teams need the same core skills everywhere and different guardrails in each country. It is easy to get this backwards by translating the slides and leaving the data rules generic.
| Stays the same everywhere | Changes by country |
|---|---|
| What AI is good and bad at | Which data law applies and what it demands |
| How to write a clear prompt | Which languages staff work in, and who can check them |
| Checking every fact, figure and reference | Which tools are approved and available on real accounts |
| The list of what never goes into a public tool | Session times and working days |
| Excel, Word and email skills | Connection quality and data cost |
| A human signs off every output | Who answers questions locally |
Regional roles also bring tasks that single-country teams rarely have:
- Adapting a group policy or HR notice for several countries without inventing local rules.
- Consolidating monthly figures in several currencies.
- Summarising a regulator's circular for colleagues in another country.
- Preparing the same customer notice in several languages.
- Writing short updates for a regional head or a board that sits elsewhere.
These make better training exercises than generic examples, because staff can check the results against work they know.
How data law differs by country
Seven countries give you six different data protection regimes and one gap. The two tables below show the position as of 8 November 2026. Laws and regulators' practice move, so have counsel in each country confirm anything you rely on.
The law, the regulator and registration
| Country | Main law and status | Regulator | Registration |
|---|---|---|---|
| Kenya | Data Protection Act, 2019. In force since 25 November 2019 | Office of the Data Protection Commissioner (ODPC) | Required. Exempt if annual turnover is below KES 5 million and staff are fewer than 10, except listed purposes such as financial services, health care, telecoms and direct marketing. Certificates last 24 months |
| Uganda | Data Protection and Privacy Act, Cap. 97, commenced 3 May 2019. Regulations took effect 12 March 2021 | Personal Data Protection Office (PDPO), under NITA-U | Required, including for entities outside Uganda handling Ugandan citizens' data. Renewed annually |
| Tanzania | Personal Data Protection Act, 2022. In force 1 May 2023 | Personal Data Protection Commission (PDPC), fully operational from 3 April 2024 | Required for all controllers and processors. Valid five years |
| Rwanda | Law No. 058/2021. In force 15 October 2021, with two years for existing operators to comply | National Cyber Security Authority (NCSA) | Required. Certificate issued within 30 days of application |
| Ethiopia | Personal Data Protection Proclamation No. 1321/2024. In force 24 July 2024 | Ethiopian Communications Authority (ECA) | Required for controllers and processors. Certificates renew every two years. A registration portal was previewed at a November 2025 workshop, and four supporting directives were reported as moving towards approval |
| Somalia | Data Protection Act, Law No. 005 of 2023. In force from March 2023 | Data Protection Authority, launched February 2024 | The Act requires "data controllers of major importance" to register within six months of qualifying (Article 32), and defines the term widely. A May 2026 Authority notice said all controllers and processors operating in Somalia must register |
| South Sudan | No dedicated law found. A Data Protection Bill was "in preparation" in September 2026. The Constitution (Article 22) protects privacy, and the Cybercrime and Computer Misuse Act, 2026 is law | No data protection regulator found | None found |
Moving data abroad, breaches and automated decisions
| Country | Sending personal data abroad | Breach notice | Decisions made solely by automation |
|---|---|---|---|
| Kenya | One of four bases: safeguards, adequacy, necessity or consent (sections 48 and 49, regulation 40). Sensitive data needs consent and confirmed safeguards. ODPC guidance published 8 September 2026 treats cloud storage abroad as a transfer and offers standard clauses, which are encouraged and not mandatory | Within 72 hours where there is a real risk of harm (section 43) | Right not to be subject to them, with exceptions (section 35) |
| Uganda | Allowed if the destination offers adequate or equivalent protection, or the person consents (section 19). No PDPO approval per transfer, but records of basis and safeguards must be available for inspection | "Immediately" (section 23) | A person can require that significant decisions are not solely automated, with exceptions (section 27) |
| Tanzania | A PDPC permit is required before transfer, including to countries with adequate protection. The application must show an international agreement, a bilateral agreement or a contract | Without undue delay, with no fixed hours (section 27(5)) | A person can require that significant decisions are not solely automated, with exceptions (section 36) |
| Rwanda | NCSA authorisation with safeguards, or consent, or listed necessity (Article 48), plus a written contract (Article 49). Data must be stored in Rwanda unless the registration certificate authorises storage abroad (Article 50) | Within 48 hours (Article 43) | A right is recorded by DataProtection.Africa (2022). Controllers must also tell people about automated decisions and the logic used (Article 42) |
| Ethiopia | Adequacy, approved safeguards or explicit informed consent. Sensitive data needs prior ECA authorisation. Tech Hive, Strathmore CIPIT and Digital Policy Alert read Article 22 as requiring storage in Ethiopia. DLA Piper's and Legal 500's summaries do not mention it, and we could not open the official text. Take local advice before choosing a cloud tool | Within 72 hours | Right not to be subject to them. High-risk processing, including profiling, requires an impact assessment |
| Somalia | Adequate protection, including contract clauses or binding corporate rules (Article 30), or consent or contract grounds (Article 31). A narrow exception for rare transfers requires telling the Authority | Within 72 hours where there is risk to individuals (Article 25) | Right not to be subject to them, with exceptions (Article 23) |
| South Sudan | No statutory transfer rule found. Follow the law where your other staff sit, your group policy and any donor clause | No statutory duty found | No statutory rule found |
Some of this comes from law firm summaries, because official texts were not always readable online. Uganda's section numbers follow the Act as passed in 2019. DLA Piper's pages were last modified between January 2025 and March 2026. The Africa AI Policy Lab recorded, with evidence to 30 June 2026, that Ethiopia's enforcement mechanisms were not yet operational. The duties are in the statute regardless, so plan for them.
What the tables mean for a training programme
Typing is moving. When someone pastes personal data into a tool hosted abroad, treat it as sending that data out of the country. Kenya's regulator treats cloud storage abroad as a transfer. Its draft AI guidance, open for comment until 17 August 2026, would expect written contracts with overseas AI vendors. We found no final version. A personal account on a free tool gives you no negotiated contract at all.
The training itself moves data. A provider in one country receiving names, emails, job titles, survey answers and recordings of staff in another is a cross-border flow. Tanzania's permit rule makes this especially worth checking. Ask counsel, and reduce what travels:
- Keep the attendance register on your side and share only headcount by site.
- Use a work email for joining links, and no ID numbers or dates of birth.
- Run the needs survey anonymously.
- Decide before day one whether sessions are recorded, where the file sits and when it is deleted.
One rule beats seven. Apply the strictest rule to any work that crosses borders. A single policy that says "no personal data in public AI tools, anywhere" is easy to teach and removes most of the country variation. Our guide to an AI usage policy was written for Kenya, and its structure transfers well.
People decide about people. Hiring, promotion, discipline, credit and benefits decisions should stay with a named human in every country, including South Sudan, where we found no statutory rule.
Where AI policy stands
Most of the region is still at strategy stage, and data protection law is what binds you today.
| Country | National AI strategy or policy | Dedicated AI law |
|---|---|---|
| Kenya | National AI Strategy 2025–2030 launched 27 March 2025. A wider AI policy is being finalised | None. The AI Bill, 2026 is a Bill. In October 2026 a Senate committee recommended not proceeding, and we found no report of a full Senate vote |
| Uganda | Under development, not yet adopted | None |
| Tanzania | Strategy in preparation, not yet adopted. Sector guidance exists, including national guidelines for AI in education | None |
| Rwanda | National AI Policy approved by Cabinet on 20 April 2023 | None. One is under consideration |
| Ethiopia | National AI Policy, 2024, endorsed at federal level (not binding) | None |
| Somalia | None found | None found |
| South Sudan | None. The communications regulator is developing AI guidelines | None |
The six rows below Kenya draw on Africa AI Policy Lab country profiles dated 7 October 2026, with evidence to 30 June 2026. Somalia's profile records a lack of verified evidence, which is weaker than a confirmed absence.
What the East African Community changes, and what it does not
For training purposes, the EAC changes very little yet. National data laws still govern.
The EAC has eight Partner States: Burundi, DR Congo, Kenya, Rwanda, Somalia, South Sudan, Tanzania and Uganda. Ethiopia is not one of them, so any plan that relies on EAC arrangements will not cover an Addis Ababa office.
Three recent developments are worth watching:
- AI declaration. At a conference in Kigali on 1 April 2026, EAC ministers adopted a declaration on AI. It calls for harmonised data governance, and for harmonised curricula, certification and accreditation in AI skills.
- Regional AI strategy. A strategy for 2026 to 2031 was reported as validated in May 2026.
- Cross-border data flows. Experts met in Dar es Salaam from 23 to 27 June 2026 to validate a proposed framework with common transfer standards. In the latest report we found, it was still a proposal, so do not plan around it.
Language is the one area where the EAC touches daily work. The November 2024 Summit approved treaty amendments adding Kiswahili and French as official languages of the Community alongside English.
Language: what AI handles and what people must own
Keep the AI step in English where accuracy matters, and give every other language a fluent human owner. Quality drops as you move away from English, and the vendors' own lists show how uneven support is.
AfroBench, a benchmark published in Findings of ACL 2025, tested models across 64 African languages. It found large gaps between high-resource languages such as English and African languages on most tasks.
| Country | Working language reality in law |
|---|---|
| Kenya | Kiswahili is the national language. Kiswahili and English are official |
| Uganda | English is official. Swahili is the second official language, for uses Parliament prescribes |
| Tanzania | A 2021 amendment made Kiswahili the language of legislation and court proceedings, with English as an exception |
| Rwanda | Ikinyarwanda, English and French are official |
| Ethiopia | Five federal working languages: Amharic, Afaan Oromo, Tigrinya, Somali and Afar. Amharic remains dominant in practice |
| Somalia | Somali (Maay and Maxaa-tiri) is official. Arabic is the second language |
| South Sudan | English is the official working language. All indigenous languages are national languages |
Here is what vendors listed when we checked on 8 November 2026.
| Language | Microsoft Copilot (text) | Gemini app | Google Cloud Translation |
|---|---|---|---|
| Kiswahili | Not listed (voice on personal accounts only) | Listed | Listed |
| Amharic | Not listed (voice on personal accounts only) | Listed | Listed |
| Somali | Not listed (voice on personal accounts only) | Not listed | Listed |
| Kinyarwanda | Not listed | Not listed | Listed |
| Luganda | Not listed | Not listed | Listed |
| Afaan Oromo | Not listed | Not listed | Listed |
| Tigrinya | Not listed | Not listed | Listed |
| Arabic | Listed | Listed | Listed |
| French | Listed | Listed | Listed |
| Juba Arabic | Not listed | Not listed | Not listed |
A language missing from a list means the vendor makes no promise about it. The tool may still reply. A language on a list tells you nothing about how well it handles your industry's vocabulary, so test it.
Three working rules follow:
- Use AI to make the English source short and clear, then hand it to a fluent colleague.
- Never publish AI output in a local language without a named fluent reviewer.
- Do not use AI to read customer complaints or staff feedback in a local language without a fluent person sampling the results.
A ten-minute exercise gives you real evidence. Give the tool a real notice and ask for a Kiswahili, Amharic or Somali version. Fluent staff mark the errors, and the team can see how far to trust the tool.
How to run one programme across countries and time zones
Run one curriculum and one policy, with mixed-country groups for the core and country breakouts for rules and language. Mixed groups let an Addis Ababa analyst and a Kampala analyst compare what works.
Time zones
The region spans two zones, and the difference catches people out.
| Zone | Countries |
|---|---|
| East Africa Time (UTC+3) | Kenya, Uganda, Tanzania, Ethiopia, Somalia |
| Central Africa Time (UTC+2) | Rwanda, South Sudan, Burundi, eastern DR Congo |
South Sudan moved to UTC+2 on 1 February 2021, so older calendars may be wrong. A session at 10:00 in Nairobi is 09:00 in Kigali and Juba.
Ethiopia needs one more line on the invitation. Many Ethiopians use a 12-hour clock that starts at sunrise, so 07:00 East Africa Time is "one o'clock" locally and 10:00 is "four o'clock". Write "10:00 EAT" in full and confirm which clock people are reading. Ethiopia's calendar also differs from the Gregorian one, so write dates out in full.
Ask each site which days are working days and which public holidays or religious observances fall in the programme period. Do not assume one calendar fits all.
A suggested eight-week shape
| Week | What happens | Who |
|---|---|---|
| Before | Written scoping: tasks, headcount, sites, devices, connection, languages | Sponsor, HR, IT |
| 1 | Regional policy agreed: approved tools, the never-paste list, human sign-off, who to ask | Legal, IT, HR |
| 2 | Test logins and tools from every country on real accounts | IT, site champions |
| 3 | Live session: what AI is, the four Cs, checking outputs | Mixed-country groups |
| 4 | Live session: Word, Excel and email on real, cleaned work | Mixed-country groups |
| 5 | Breakouts: each country's data rules and language testing | By country |
| 6 | Live session: function-specific tasks (finance, HR, sales, operations) | By function |
| 7 | Practice on a real task, with champions holding an open clinic | Everyone |
| 8 | Review: time saved, errors caught, policy updates | Sponsor, champions |
Keep live sessions to 60–90 minutes and groups small enough for everyone to practise. Send a short written recap after each one. A programme spread over weeks gives people time to build habits, as we explain in why corporate AI training does not stick.
Name two champions per site, from different teams. They answer questions after the trainer has gone.
Design for the weakest connection
Internet access varies more than most regional plans allow for. DataReportal's figures for October 2025 show internet use as a share of population:
| Kenya | Rwanda | Tanzania | Somalia | Uganda | Ethiopia | South Sudan |
|---|---|---|---|---|---|---|
| 40.5% | 34.2% | 29.1% | 27.6% | 22.0% | 21.7% | 13.2% |
Office staff are likely to be better connected than these national averages, but field, branch and warehouse staff may not be. Plan for that:
- Send a PDF pack of prompts and exercises before the session, so everything survives a dropped call.
- Keep cameras optional and demonstrations short.
- Teach every exercise on a phone as well as a laptop.
- Avoid building the course around long video.
Check the political calendar too. Access Now reports five-day shutdowns during Tanzania's October 2025 election period and Uganda's January 2026 election, and at least 30 shutdowns in Ethiopia since 2016. It listed South Sudan's general election for 22 December 2026, with its status uncertain. Keep important sessions away from election dates, and make sure staff hold the materials offline.
Check the tools from each country
Availability lists are a starting point, so test real accounts from each country. As of 8 November 2026:
| Tool | What the vendor says | Before staff use it |
|---|---|---|
| Gemini | The app lists all seven countries | Google says some chats go to human reviewers and warns against entering confidential information. Teach the Keep Activity setting |
| Claude | All seven are supported countries | Consumer chats can be used for training if that setting is on. Work plans are excluded |
| ChatGPT | OpenAI's API list includes all seven. Test the app on real accounts | Personal accounts can be used for training unless switched off in Data Controls. OpenAI says Business and Enterprise workspaces are not used by default |
| Microsoft 365 Copilot Chat | Available in over 170 markets, with embargoed markets excluded. The page names no East African country | Needs a qualifying Microsoft 365 business or enterprise subscription. Without a Copilot licence, work data is limited to files you upload |
If your company runs Microsoft 365, give Word, Excel and Copilot Chat real time in the course. Show the free version of each feature, and let IT decide what is approved.
What goes wrong, and how to avoid it
The problems we would watch for most come from skipping the policy or the local layer.
| What goes wrong | How to avoid it |
|---|---|
| Head office writes the policy and countries ignore it | Have each country's HR or legal lead review it first |
| A Kenyan policy is copied to Tanzania or Rwanda unchanged | Use the tables above and ask local counsel |
| Staff paste payroll data to "tidy it up" | Teach the never-paste list on day one, and offer a safe alternative |
| AI invents a local law or a deadline | Paste the official text and tell it to use only that |
| Only head office is trained | Put branch and field staff in the first groups |
| Licences are bought before anyone has tested the use | Run the programme on existing tools, then decide |
| Success is reported as "people enjoyed it" | Time one task before and after, including checking time |
Training has limits. It will not fix a site with no devices or data budget. It will not settle a legal question about your transfers. And it should wait if leadership has not agreed who may use which tools.
Seven sample prompts for regional teams
Each prompt follows the four Cs we teach: Clarity, Context, Constraints and Customisation. Replace the square brackets, and keep personal data out.
1. Adapting a group notice for one country
Rewrite the group HR notice below for staff in [country]. I work in regional HR at a company with offices in [countries]. Keep every date and rule exactly as written. Do not add or assume anything about local law. Where the notice may depend on local employment law or public holidays, write [CHECK WITH COUNTRY HR]. Maximum 200 words, warm and clear. Readers will skim it on a phone. Notice: [paste with no personal details]
2. Comparing two regulators' published texts
Compare the two published texts below on [topic] in a table with columns: Rule, Section number, What it requires, Deadline. Use only the text I paste. Do not use your general knowledge of these laws. Where a text is silent, write "not stated". I am preparing a briefing for a regional compliance lead, who will confirm it with counsel. Text A: [paste]. Text B: [paste]
3. An invitation that survives two time zones
Write a short training invitation for staff in Nairobi, Kampala, Dar es Salaam, Addis Ababa, Mogadishu, Kigali and Juba. The session starts at 10:00 East Africa Time on [date] and lasts 90 minutes. Use UTC+3 for the first five cities and UTC+2 for Kigali and Juba, and show the local start time in a table. Add one line telling Addis Ababa colleagues this is East Africa Time on the international clock. Maximum 120 words, friendly.
4. Consolidating figures in several currencies
I have an Excel sheet with columns: Country, Entity, Currency, Local Amount. Currencies are KES, UGX, TZS and RWF. A second sheet called Rates has Currency in column A and the rate to USD in column B. Give me a formula to convert each row to USD, a formula to flag any currency missing from Rates, and a summary by country. Explain each formula in one sentence. I use basic Excel and have not used pivot tables.
The figures stay in your sheet, because the tool only needs the column names.
5. Preparing text for a human translator
Rewrite this customer notice in very simple English so that a colleague can translate it into [Kiswahili, Amharic, Somali or another language]. Use short sentences, one idea each, and no idioms. Keep the date, time and place exactly as written. Then list any words likely to be hard to translate. Notice: [paste]
6. Cleaning a document before using AI
I will not paste the document. It is a [staff performance summary] from our [country] office. List the kinds of personal or confidential detail it is likely to contain, and suggest a neutral replacement label for each, such as [Staff A] or [Branch 2]. Present this as a checklist I can follow by hand.
7. Checking the draft
List every factual claim, number, date, name and reference in the text below. For each one, say whether it appears in the source I have pasted. Write "not in source" where it does not. Do not correct anything yet. Text: [draft]. Source: [source]
Prompt 7 is the one to teach first. It protects people in every country and every department.
A checklist for choosing a provider
Ask these questions before you sign, and expect specific answers. Our procurement checklist and our guide on how to choose a provider go further.
| Ask | A good answer | A warning sign |
|---|---|---|
| Which of our countries have you delivered in? | Names them honestly, and says where it has none | "We cover all of Africa" with no detail |
| How do you handle each country's data law? | Refers to the law and suggests counsel confirms | One slide on GDPR |
| What staff data do you need from us? | As little as possible, with a deletion date | Asks for full staff lists by default |
| Which languages do you teach in? | States them, and explains how local output gets checked | Claims AI handles every language well |
| Do you favour one tool? | Works with what you have approved | Sells licences or one product |
| How do you handle two time zones and poor connections? | Offers a concrete schedule and an offline pack | Sends long recordings |
| Will you use our real tasks? | Scopes them with you in writing first | Same deck for every client |
| What happens after the last session? | Champions, written recaps, a review point | Certificate and goodbye |
| How will we measure results? | Time a real task and log errors | Promises a percentage gain |
| Is anything claimed about accreditation? | Shows the certificate, and is clear where it applies | Vague references to "accredited" courses |
| Which entity contracts and invoices? | Agrees this early so your finance team can check tax on a cross-border service | Leaves it until the invoice |
Questions people ask
Can one AI training programme work across several East African countries?
Yes, if the core is shared and the data rules and language work are local. A single policy built on the strictest rule keeps it manageable.
Which East African country has the strictest rules on moving data abroad?
Tanzania and Rwanda stand out. Tanzania requires a PDPC permit before transfer. Rwanda's default is storage in Rwanda unless the registration certificate authorises otherwise. Ethiopia also restricts transfers, and several summaries read its law as requiring local storage.
Is there one data protection law for the whole East African Community?
No. Each country's own law applies. The EAC was validating a proposed cross-border data flows framework in June 2026, and we found no report of it being finalised. Ethiopia is outside the EAC altogether.
Can staff use ChatGPT or Copilot with customer data?
Personal data should stay out of public tools in every country. Work-managed tools have stronger terms, but you still need a lawful basis and any transfer steps your country requires. Ask your data protection lead before allowing it.
Can AI write well in Kiswahili, Amharic or Somali?
It can produce a usable draft, and quality varies by tool and subject. Research such as AfroBench finds large gaps between English and African languages. A fluent colleague should review anything that reaches customers or staff.
Should we train online or in person?
Online suits teams spread across countries and keeps costs down. In person suits a group sharing a room with a poor connection. A blend can work well: live online core sessions, with an in-person start at the largest site.
How long does regional AI training take?
Plan about eight weeks, including policy and testing, with four live sessions and a practice clinic. Nobody becomes an expert in a few days, so spread the practice out.
Sources
All sources last checked 8 November 2026.
Data protection law
- Kenya, Data Protection Act, 2019 (ODPC copy)
- DLA Piper, data protection in Kenya (modified 23 March 2026)
- Afriwise, "Data Without Borders? What Kenya's New Cross-Border Data Transfer Guidance Means for Businesses" (28 October 2026)
- Business Daily, "Regulator now plans tighter data shields on offshore AI platforms" (4 August 2026)
- Digital Policy Alert, ODPC consultation on its draft AI guidance note, to 17 August 2026
- Uganda, Data Protection and Privacy Act, 2019 (Parliament copy)
- DLA Piper, data protection in Uganda (modified 13 February 2026)
- Afriwise, on the PDPO's decision concerning offshore entities (Ssekamwa Frank and 3 others v Google LLC, 18 July 2025)
- Tanzania, Personal Data Protection Act (PDPC copy)
- DLA Piper, data protection in Tanzania (modified 24 March 2026)
- Clyde & Co, cross-border personal data transfers in Tanzania (October 2024)
- DLA Piper, data protection in Rwanda (modified 6 January 2025)
- DataProtection.Africa, Rwanda fact sheet (updated 23 May 2022)
- DLA Piper, data protection in Ethiopia (modified 23 March 2026)
- Digital Policy Alert, Ethiopia's data localisation requirement
- Tech Hive Advisory, review of Ethiopia's data protection law
- Strathmore CIPIT, "Ethiopia's Personal Data Protection Proclamation of 2024" (12 November 2024)
- Legal 500, Ethiopian personal data protection compliance guide (30 March 2026)
- Addis Fortune, "ECA launches digital portal to enforce data protection"
- Somalia, Data Protection Act, Law No. 005 of 2023
- Tech Hive Advisory, review of Somalia's Data Protection Act and implementing regulations
- Dawan Africa, Somali Data Protection Authority's registration notice (31 May 2026)
- KICTANet, "South Sudan Ties Data Rules to Citizen Trust" (11 September 2026)
- Eye Radio, "Controversial cybercrime act takes effect amid free expression concerns"
AI policy and the EAC
- Africa AI Policy Lab, Uganda profile (7 October 2026)
- Africa AI Policy Lab, Tanzania profile (7 October 2026)
- Africa AI Policy Lab, Rwanda profile (7 October 2026)
- Africa AI Policy Lab, Ethiopia profile (7 October 2026)
- Africa AI Policy Lab, Somalia profile (7 October 2026)
- Africa AI Policy Lab, South Sudan profile (7 October 2026)
- Citizen Digital, Kenya's National AI Strategy launch
- The Star, "Senate committee shoots down Nyamu's Bill on AI" (15 October 2026)
- East African Community press release naming the eight Partner States (6 May 2026)
- Tech Africa News, EAC ministers' AI declaration (6 May 2026)
- ITWeb Africa, "East Africa validates regional AI strategy" (6 May 2026)
- Tech Africa News, EAC cross-border data flows framework (3 July 2026)
- Citizen Digital, "EAC adds Kiswahili and French as official languages"
Languages
- Constitution of Kenya, Article 7
- Constitution of Uganda, Article 6
- Constitution of Rwanda, Article 8
- Provisional Constitution of Somalia, Article 5
- Transitional Constitution of South Sudan, Articles 6 and 22
- Afriwise, Kiswahili in Tanzanian law and courts
- LSE Africa at LSE, on Ethiopia's working languages (7 July 2023)
- Ojo et al., "AfroBench: How Good are Large Language Models on African Languages?", Findings of ACL 2025
- Microsoft, supported languages for Copilot
- Google Cloud, Translation language support
Time, connectivity and shutdowns
- IANA Time Zone Database, Africa
- IANA Time Zone Database, city links
- The EastAfrican, South Sudan's time zone change (29 January 2021)
- World Travel Guide, Ethiopia business communications
- DataReportal, Digital 2026: Kenya
- DataReportal, Digital 2026: Uganda
- DataReportal, Digital 2026: Tanzania
- DataReportal, Digital 2026: Rwanda
- DataReportal, Digital 2026: Ethiopia
- DataReportal, Digital 2026: Somalia
- DataReportal, Digital 2026: South Sudan
- Access Now, 2026 elections and internet shutdowns watch
AI tools
- Google, where the Gemini app is available and supported languages
- Google, Gemini Apps Privacy Hub
- Anthropic, supported countries
- Anthropic, updates to consumer terms
- OpenAI, API supported countries
- OpenAI, how your data is used to improve model performance
- Microsoft, supported regions for Copilot
- Microsoft, Copilot Chat with and without a licence
Where FuKazee fits in
FuKazee is a training company based in Nairobi, and that is our only address. We have trained more than 3,000 people in tech, the future of work and everyday AI. We make no claim here of offices, clients or results in the other countries this guide covers. Ask us directly what we have and have not done, as you should ask any provider.
Training for regional teams is available live online, or in person by arrangement. We teach in English and rely on your fluent staff for the language checks described above. We do not give legal advice, so your counsel in each country stays part of the picture.
We assume nobody in the room knows AI yet, and we avoid jargon. We show the free version of every feature and do not promote one tool. Microsoft 365 and Copilot are part of the course, and so are the four Cs, safe data handling and checking every output. Nobody becomes an expert in a few days, so we build practice in over several weeks. You can read more about how we work with organisations on our page for AI training in Kenya.
To discuss a regional programme, tell us your countries, how many people, what they do and what you want AI to help with. Email lead@fukazee.com.
Want guided support instead?
Plan AI training around your team's real work. Answer a few quick questions about your sector, team size and goals, and we will recommend the right programme and format. Sessions are built on your workflows and include a usage policy.
Rodney has run FuKazee's training since 2022 and still teaches inside every programme line: corporate sessions, the postgraduate cohort and the KCSE examiner webinars. First Class Honours in Journalism and Mass Communication, University of Nairobi.
