How to protect your privacy when using AI tools

AI tools such as ChatGPT, Gemini, Claude, Copilot and Meta AI inside WhatsApp have quickly become part of daily life in Uganda. Students use them for revision, accountants use them to draft reports, and shop owners use them to write adverts. But every prompt you type is information you are handing to a company. Knowing how to protect your privacy when using AI tools is now as important as knowing how to use them well. This guide gives you clear habits and settings to keep your personal and business information safe.

Why privacy matters with AI tools

Most AI chat services process your conversations on their own servers, often outside Uganda. Depending on the service and your settings, conversations may be:

• Stored in your account history for a period of time.
• Used to improve or train future models.
• Reviewed by staff or contractors for safety and quality checks.
• Kept for legal or security reasons even after you delete them.

None of this is necessarily sinister, and policies differ between providers. The point is simple: treat anything you type into an AI tool as something that could be seen by someone else.

The golden rule: do not share what you would not post publicly

Before you paste anything into an AI chat, ask yourself: "Would I be comfortable if this appeared on a notice board?" If not, remove it or rewrite it. Information you should never share includes:

• National ID numbers, passport numbers and dates of birth.
• Bank account numbers, card details and Mobile Money PINs.
• Passwords, one time codes and API keys.
• Medical records or details about someone's health.
• Client lists, payroll files, contracts or confidential tender documents.
• Photos of other people, especially children, without their consent.

How to anonymise information before using AI

You can still get useful help from AI without exposing real data. The trick is to strip out identifying details first.

1. Replace names with roles: "Client A", "the supplier", "Staff member 1".
2. Remove contact details such as phone numbers, emails and physical addresses.
3. Round or change numbers where the exact figure is not needed. "About UGX 5 million" is often enough.
4. Summarise instead of pasting. Rather than uploading a whole HR file, describe the situation in general terms.
5. Use sample data for spreadsheet formulas. Create a small fake table with the same columns as your real one.

For example, instead of "Write a warning letter to Sarah Namubiru, employee number 0457, who was late seven times in March at our Kireka branch", try "Write a polite warning letter to an employee who has been late several times this month." You will get an equally good letter and can add the details yourself.

Check and adjust your settings

Most major AI services now offer privacy controls. They move around as apps are updated, so look in Settings, Privacy, or Data Controls. Common options include:

• Turning off model training on your conversations.
• Temporary or incognito chats that are not saved to your history.
• Deleting chat history individually or all at once.
• Controlling memory features that let the assistant remember facts about you between chats.
• Managing connected apps such as email, calendar or cloud storage.

Spend ten minutes going through these settings on each AI tool you use. Repeat every few months, because new features are often switched on by default.

Be careful with AI that connects to your accounts

Many assistants can now read your email, documents or calendar. This is powerful but raises the stakes. Only connect what you genuinely need, review which permissions you are granting, and disconnect services you stopped using. On Google, Microsoft and other accounts you can view connected third party apps in the security section and remove access.

Watch out for fake AI apps and scams

The popularity of AI has created a wave of copycat apps and scams. Protect yourself by:

• Downloading AI apps only from the official Google Play Store or Apple App Store, and checking the developer name.
• Avoiding browser extensions that promise free premium AI, as some steal passwords or browsing data.
• Ignoring WhatsApp or Facebook messages offering paid AI accounts at suspicious prices.
• Being sceptical of voice notes or videos that sound like someone you know asking for money. AI can now clone voices, so confirm by calling the person directly.

Privacy tips for businesses and organisations

If your staff use AI tools at work, you need some ground rules. Uganda's Data Protection and Privacy Act places duties on organisations that collect and process personal data, and the Personal Data Protection Office oversees compliance. Pasting customer data into an outside tool without care could create problems. A simple AI policy might cover:

1. Which AI tools are approved for work use.
2. What types of information must never be entered.
3. Whether business or team plans with stronger privacy terms should be used.
4. Who to ask when unsure.
5. A requirement to check AI output for accuracy before sending it to clients.

Business versions of AI services often come with clearer commitments about not training on your data. Read the terms, or ask your IT provider to explain them.

Protect the device and account too

Your AI chat history is only as safe as the account it sits in. Basic security still applies:

• Use a strong, unique password for each AI account.
• Turn on two step verification.
• Lock your phone with a PIN or fingerprint, especially if you share it with family.
• Sign out of AI tools on shared computers at cyber cafés, libraries or the office.

Quick privacy checklist

• Remove names, numbers and IDs before prompting.
• Turn off training on your data where possible.
• Use temporary chats for sensitive questions.
• Review and delete old history regularly.
• Limit connected apps and permissions.
• Download only official apps.
• Agree workplace rules for AI use.

Frequently asked questions

Does deleting a chat remove it completely?

Not always immediately. Many providers remove deleted chats from your view straight away but keep them on their systems for a limited period. Check the provider's privacy policy for details.

Is it safe to upload documents to AI tools?

It can be, for non sensitive material. For contracts, financial records or personal data, remove identifying details first or use an approved business tool.

Are free AI tools less private than paid ones?

Not automatically, but paid business plans often offer stronger privacy commitments. Whatever you use, read the settings rather than assuming.

Can AI tools see my other phone apps?

Generally only if you give permission, for example by connecting accounts or allowing access to files. Review app permissions in your phone settings.

Conclusion

AI tools can save you hours, but they work best when you stay in control of your information. Share less, anonymise more, check your settings and keep your accounts secure. With these habits you can enjoy the benefits of AI without putting your privacy, or your clients' trust, at risk. For more practical tech guides, keep following the Baasa Cloud blog.