Think Before You Share: What Kenyans Should Never Tell AI Chatbots

From asking artificial intelligence (AI) chatbots about health concerns to seeking relationship advice, planning meals and drafting personal messages, people are increasingly trusting the technology with information they might previously have shared only with friends, professionals or family members.

But cybersecurity experts are warning that this growing reliance on AI could expose users to privacy breaches, misinformation and online scams if they share sensitive information without understanding how the services handle their data.

A survey by cybersecurity firm Kaspersky found that nearly 79 per cent of respondents across the Middle East, Turkiye and Africa (META) region use AI services, including chatbots and image and video generation tools.

The findings highlight how AI is moving beyond workplace productivity and entertainment into people’s personal lives, raising questions about what happens to the information users disclose to these platforms.

According to the survey, 74 per cent of respondents use AI to search for or summarise information, while half use it to write texts. Others use the technology to generate images, edit photographs and create videos.

More significantly, 41 per cent said they seek advice on personal issues, while 37 per cent turn to AI for health-related questions. Another 29 per cent use it to plan leisure activities, including holidays and parties, while 23 per cent rely on it to prepare meal plans.

For Kenyan users, the trend presents a growing digital safety concern as AI tools become more accessible through smartphones and online platforms.

A person seeking advice about a medical condition, for instance, might disclose symptoms, medication details or other private information. Someone experiencing relationship difficulties might share intimate conversations, names or personal circumstances while seeking guidance from a chatbot.

A small business owner might also upload customer records or financial documents to obtain a summary or analysis without considering whether the information should be shared with an external service.

Although AI platforms differ in how they collect, retain and process information, users should understand their privacy policies before submitting sensitive material. Depending on the service and its settings, conversations may be stored or reviewed, and some information may be used to improve AI systems.

Information you should never share with AI chatbots

Cybersecurity experts advise users to keep the following information out of AI conversations:

• Passwords and login credentials: Never share passwords, personal identification numbers (PINs), one-time verification codes, account recovery codes or security-question answers. These could expose email, social media, banking or mobile money accounts.
• Banking and mobile money details: Avoid entering bank account credentials, payment card numbers, card security codes or M-Pesa PINs that could enable unauthorised transactions.
• Personal identification documents: Do not upload copies of national identity cards, passports, driving licences or birth certificates containing sensitive identifying details unless there is a clear, necessary and appropriately protected reason.
• Private health information: Avoid sharing medical records, laboratory results, prescriptions, hospital numbers or detailed medical histories alongside your name and other identifying information. You can ask general health questions without revealing your identity.
• Home addresses and live locations: Do not unnecessarily disclose your precise home address, real-time location, daily movements or travel plans, particularly when combined with your name and contact details.
• Confidential work and business information: Do not upload internal company reports, customer databases, patient records, unpublished research, financial statements or confidential contracts without proper authorisation and safeguards.
• Private conversations and intimate images: Be cautious about uploading private chats, intimate photographs or personal recordings involving yourself or others. Consider consent and privacy before sharing another person’s information.
• Children’s personal details: Avoid disclosing children’s full names, school details, photographs, home addresses or medical records when these details are not necessary.
• Security-sensitive information: Keep alarm codes, workplace access codes and confidential security procedures private because they could help someone gain unauthorised access to your home, workplace or accounts.

The guiding principle is to share only the information necessary to complete a task. For example, someone seeking help drafting a medical appointment request can describe the general concern without providing a national identification number, hospital number or full medical history.

Kaspersky warns that information shared with AI services could be exposed or misused, depending on the platform’s security, data-handling practices and the circumstances under which third parties may access it.

The reliability of AI-generated information is another concern.

The survey found that nearly one in 10 respondents had encountered incorrect information generated by AI, while 24 per cent believed AI could provide inaccurate answers.

This raises particular concerns when users seek medical, financial or legal guidance and act on responses without consulting qualified professionals or checking reliable sources. AI chatbots can produce convincing answers that contain factual errors, outdated information or misleading recommendations.

“Users should be aware of the limitations and potential risks associated with AI services,” said Seifallah Jedidi, Kaspersky’s Head of Consumer Channel in the Middle East, Turkiye and Africa.

He urged users to verify information, handle personal data cautiously and assess AI-generated responses critically. He also warned about fraudulent services that imitate legitimate AI platforms to distribute malicious software.

Fake AI applications and websites may trick users into revealing login credentials, downloading harmful software or paying for services that do not deliver what they promise. Users should therefore be cautious when downloading unfamiliar applications or following links advertising AI services.

Reviewing privacy policies and account settings can help users understand how their conversations may be collected, stored or used. Where available, privacy controls can limit how information is retained or used to improve AI systems, although the options differ between platforms.

As AI becomes part of everyday life, the challenge is not simply learning how to use it but understanding what information should never be shared with it.

For Kenyan consumers, a useful rule is to treat an AI chatbot as a digital tool rather than a confidential adviser whose responses are automatically accurate or whose handling of personal information can be taken for granted. Before typing, users should ask themselves whether they would be comfortable with the information being stored, accessed or exposed.