6. Using AI safely and protecting your privacy

When you type a message to an AI chatbot, that message travels over the internet to the company's servers where it is processed. Most major AI companies store your conversations — at least for a period of time — to improve their systems. This means what you type is not completely private, the same way a phone call or text message is not fully private.

This does not mean AI is dangerous. It means you should treat AI conversations the way you would treat a conversation in a public place: comfortable for general topics, but not the right venue for your deepest secrets, passwords, or ID numbers.

Abebe works at a small electronics shop near Merkato. He uses Claude to write customer messages and marketing ideas. That is perfectly fine — no sensitive personal information is involved. But if Abebe typed his shop's bank account password into Claude to 'ask it to check the balance,' that would be a serious mistake.

Here is a clear list of information you should never type into an AI chatbot:

• Passwords — for your phone, bank, email, TeleBirr, social media, or any other account.

• PINs — for your ATM card, mobile money (*127# USSD PIN), or Telebirr account.

• National ID number — your Fayda ID, kebele ID number, or passport number.

• Bank account numbers and financial details — your account number, full card number, or credit details.

• Medical details — specific diagnoses or medications that you want kept confidential.

• Other people's private information — names, phone numbers, or ID numbers belonging to people who have not given you permission to share them.

A good rule to remember: if you would not write this information on a piece of paper and hand it to a stranger at a bus station in Addis Ababa, do not type it into an AI chatbot.

Not every app or website that claims to be an 'AI assistant' is safe. Scammers create fake apps that look like ChatGPT or Claude but are actually designed to steal your information or money. Here is how to stay safe:

1. Use only the official websites — ChatGPT is at chat.openai.com, Claude is at claude.ai, and Gemini is at gemini.google.com. Be careful of websites with similar-sounding names like 'chatgpt-free-africa.com' or 'claude-ai-app.net'.

2. Download apps only from official app stores — Google Play Store or Apple App Store. Even there, check that the developer name matches the real company (OpenAI, Anthropic, Google).

3. Never pay to 'unlock' a free AI tool via a strange payment link — the free versions of ChatGPT, Claude, and Gemini are available directly on their official websites at no charge.

4. If an 'AI' app asks for your Telebirr PIN, bank password, or national ID to 'verify your identity,' stop immediately. Legitimate AI tools do not ask for these.

Almaz received a WhatsApp message saying 'Free AI tool — send 50 birr to activate.' She ignored it. Smart move — this is a common scam in Ethiopia and across Africa.

Your privacy matters — and so does the privacy of the people around you. A common mistake is sharing other people's personal information with AI without thinking about it.

For example:

• Sara manages a small NGO office in Addis Ababa. She wants AI to help summarise a report about staff performance. If she pastes full names, salaries, and personal evaluations into ChatGPT, she is sharing her employees' confidential information with a third-party server without their knowledge or consent.

• Dawit is a teacher. He wants AI to help write feedback for a struggling student. If he types the student's full name, family situation, and health details, he may be violating that student's privacy.

The fix is simple: anonymise before you share. Replace real names with 'Employee A' or 'Student 1.' Remove phone numbers, addresses, and ID numbers. Describe situations in general terms rather than giving full personal details. The AI can still help you — without the privacy risk.

AI can sometimes produce text that sounds confident and accurate but is actually false or misleading. When you share that text with others without checking it, you can accidentally spread misinformation.

Here are practical steps to avoid this:

1. Verify before you share — If AI writes an article, fact sheet, or social media post for you, read it carefully. Check any specific claims — dates, statistics, names, or medical facts — against a trusted source before posting.

2. Don't use AI to impersonate people — Some people use AI to generate fake quotes from politicians, teachers, or community leaders. This is dishonest and can cause serious harm. Never present AI-generated text as if a real person said it.

3. Label AI-generated content when appropriate — If you use AI to help write a newsletter or report, it is good practice to mention that AI was used. This builds trust with your audience.

4. Be especially careful with health information — In Ethiopia, health misinformation can spread quickly through WhatsApp and Facebook. If AI gives you health advice, confirm it with a qualified health worker before sharing with family or community groups.