AI warning: 10 things you should never share with ChatGPT and other chatbots

Artificial intelligence chatbots like ChatGPT are rapidly changing the way people interact with technology. From answering quick questions and drafting emails to offering companionship and emotional support, these tools are becoming a daily presence in many lives. Their ease of use and ability to generate human-like responses make them feel reliable and even trustworthy. However, this sense of comfort can be misleading. Experts warn that

oversharing with AI

carries serious risks, including privacy breaches, identity theft, and misuse of personal or sensitive information. Unlike a human conversation, interactions with AI are not confidential, and what you share may be stored, analysed, or even exposed. Here are 10 things you must never share.

10 things that you should never discuss with AI chatbots

1. Personal information

Details like your full name, home address, phone number, or email may seem harmless but can be pieced together to track your online identity. Once exposed, these identifiers could be used for scams, targeted phishing, or even physical tracking. AI systems are not designed to provide anonymity, so keeping this information private is essential.

2. Financial details

Bank account numbers, credit card details, or Social Security numbers are prime targets for cybercriminals. If entered into a chatbot, this data could be intercepted, stored, or misused, leaving you vulnerable to fraud and identity theft. Financial information should only ever be shared through secure, official channels—not AI assistants.

3. Passwords

No chatbot should ever be entrusted with your login credentials. Sharing passwords, even in casual conversation, can put your email, banking, and social media accounts at risk. Cybersecurity experts stress that you should store passwords only in secure password managers, never in AI chats.

4. Secrets or confessions

While some may feel comfortable “venting” to a chatbot, it is not the same as confiding in a friend or therapist. AI cannot guarantee privacy, and anything you type may be logged for training or monitoring. Personal secrets or confessions could resurface or be exposed unintentionally.

5. Health or medical information

It’s tempting to ask chatbots about symptoms or treatments, but AI is not a licensed medical professional. Misdiagnoses could occur, and sharing personal health data—including medical records, prescriptions, or insurance numbers—creates risks if the information leaks. Always consult a qualified doctor for health matters.

6. Explicit or inappropriate content

AI platforms typically flag or block explicit material, but what you share may still be recorded. This includes sexual content, offensive remarks, or illegal material. Not only can such interactions result in account suspension, but sensitive data might linger in system logs, raising future risks.

7. Work-related confidential data

Companies are increasingly warning employees not to paste confidential documents or plans into AI systems. Business strategies, internal reports, or trade secrets could accidentally leak outside your organization. Since some chatbots use inputs to improve their models, sharing work-sensitive content could jeopardize corporate security.

8. Legal issues

It may be tempting to ask a chatbot for help with contracts, lawsuits, or personal disputes. However, AI cannot replace a lawyer and may provide misleading or incomplete guidance. Worse, sharing legal details about cases, agreements, or disputes could harm your legal standing if the information is ever exposed.

9. Sensitive images or documents

Never upload IDs, passports, driver’s licenses, or private photos to a chatbot. Even if deleted, digital traces can remain. Sensitive files could be hacked, repurposed, or even used for identity theft. Always keep personal documentation offline or in secure, encrypted storage.

10. Anything you don’t want public

The golden rule: if you wouldn’t want it broadcast online, don’t share it with an AI. While chatbots may seem private, most are not designed for secrecy. Even harmless comments could be logged and resurfaced in ways you cannot control. Treat every AI interaction as though it might one day be public.