FRAUD ALERT

The July 'cockroach' protests at Jantar Mantar in New Delhi brought forward two important outlooks. First, students and youngsters who make up 'Gen Z' have emerged as a very strong voice for all who are oppressed. The second, and most important, is Gen Z bypassing legacy media (read: TV channels) and using witty, humorous content to gain public support and challenge state narratives on Instagram (Insta). The thousands of reels on Insta also revealed the amazing creativity and friendliness this younger generation shows while using artificial intelligence (AI) tools.

 

While it feels really good to see emerging citizens using digital tools to amplify their voices, I am a little worried about Gen Z's overreliance on AI tools and the lack of safety and security they follow. And I am not the only one who feels this way. 

 

Research by cybersecurity and digital privacy company Kaspersky found that 32% of Gen Z respondents surveyed in India said they use AI as a friend or Pookie (a playful, affectionate nickname often used between friends or partners by Gen Z). More worryingly, from a privacy perspective, 37% said they use AI to discuss personal issues they would rather not share with others.

 

That may seem harmless. After all, there is no awkward silence, no judgement and no friend who might repeat what was said.

 

But an AI chatbot is not a human confidant. It should never be treated as a private diary. “When one in three Gen Z users in Asia-Pacific countries (APAC) already sees AI as a friend, it is clearly becoming more than just a tool. As that relationship becomes more personal, knowing what to trust, what to share and where to draw the line will be key to using AI safely and responsibly,” said Adrian Hia, managing director (MD) of Kaspersky for the Asia Pacific region. 

 

The Indian numbers from the survey point to a generation that is becoming very quickly comfortable with AI. For many young Indians, asking an AI chatbot a question is now as routine as searching Google, sending a message or opening Instagram.

 

It can help write an assignment, explain a difficult topic, suggest ideas, prepare for an interview or even offer advice when someone is struggling with a personal problem. And that last use is where a new privacy risk is emerging.

 

People are beginning to treat AI systems less like software and more like someone they can talk to. The danger is that familiarity can easily turn into overconfidence.

 

The simplest rule for using AI safely is also the one many users are likely to ignore: if you would not share the information publicly or with a stranger, do not put it into an AI chatbot. This becomes particularly important when a conversation turns personal.

 

A student may paste a private conversation and ask an AI tool to analyse a relationship. Someone facing a financial problem may upload a bank statement and ask for help understanding expenses. A job seeker may paste an entire CV containing their address and phone number. Someone involved in a dispute may upload legal documents. Another user may share medical reports because they want a simple explanation.

 

Passwords, personal identification numbers (PINs), one-time passwords (OTPs), bank account details, card information, Aadhaar and PAN details, passport information, private photographs, confidential work documents and personal messages should never be handed over casually to an AI service. It is the same caution that we apply to sharing information with other people.

 

All of these actions can expose far more information than the user intended to reveal.

 

A user who copies a friend's private messages into a chatbot for analysis is not sharing only their own information. They may also be exposing someone else's name, phone number, personal conversation or other private details.

 

Privacy is only half the problem with AI tools. The other is trust. 

 

AI tools can produce answers that sound confident and convincing even when they are inaccurate or completely wrong. This is particularly dangerous when the subject involves money, health, law, education or important personal decisions.

 

Just a few days ago, the Supreme Court set aside a ₹425.28 crore customs penalty imposed on a diamond trader after finding that the adjudicating officer had relied on non-existent cases, fake citations and legal propositions hallucinated by AI.

 

Imagine someone asking an AI chatbot whether a particular investment is safe and acting on the response without checking elsewhere. Or a student using an incorrect explanation in an examination. Someone could also rely on an AI-generated interpretation of a legal notice or medical report when professional advice is actually needed. Also, the output from AI may not give you the complete picture or may produce an imaginary (read: hallucinated) response. 

 

The conversational style of AI makes this problem harder to notice, and that can create a false sense of trust. 

 

A search engine generally feels like a search engine. A chatbot, by contrast, can respond in a warm, conversational manner, remember the context of a discussion within a session, and appear to understand what a user is going through.

 

However, remember that a machine producing a sympathetic response does not mean the conversation is confidential, the advice is correct, or the system understands the user's circumstances as a human professional would.

 

A global clinician survey published in medRxiv found 91.8% had encountered medical AI hallucinations, and 84.7% believed they could directly cause patient harm. 

 

 

I am not saying that everyone should stop using AI tools. AI can be extremely useful for education, work and everyday tasks. What needs to change is the boundary around what you are sharing.

 

Before typing anything into an AI tool, ask one simple question: “Would I be comfortable if this information were exposed?”

 

If the answer is no, leave it out.

 

For example, instead of uploading an entire bank statement, remove the account number, address and transaction identifiers and ask about a generic financial problem. Instead of pasting an entire private conversation, describe the situation without names, phone numbers or identifying details.

 

The same principle applies to documents. If an AI tool is being used to summarise a document, remove unnecessary personal information first. A document does not become safe simply because it is being uploaded to a familiar-looking chatbot. And we simply have no idea whether it would be used to train newer AI modules or not. 

 

No legitimate AI assistant needs your banking password, UPI PIN, card PIN or one-time password. Anyone asking for such information should be treated as suspicious.

 

Do not upload Aadhaar, PAN cards, passports, driving licences or other identity documents unless there is a genuine and clearly understood reason to use an authorised service. Even then, check exactly who operates the service and what happens to the information.

 

If you need AI to analyse a document, first delete information that is not necessary for the task. Names, phone numbers, addresses, account numbers, signatures and other identifying details should not be included unless essential.

 

An AI chatbot does not need to see an entire WhatsApp conversation, email chain or private message to help you draft a response. Summarise the situation and remove names and other identifying details.

 

Use AI to understand concepts or prepare questions, but do not blindly follow its recommendations about investments, loans, taxes or money transfers. Verify important information with official sources or qualified professionals.

 

For anything involving health, law, education, employment or money, check the answer against reliable sources. A fluent answer is not necessarily a correct answer.

 

Do not download an AI application simply because an advertisement, social media post or message claims it is the latest version of a popular tool. Fraudsters can create fake websites and applications designed to steal passwords, financial information or other personal data.

 

There is another danger that users should not overlook. As AI becomes more popular, cybercriminals are using its popularity as bait.

 

Messages offering free access to premium AI tools, fake AI-powered investment platforms, unofficial chatbot applications and supposed updates can lead users to malicious websites or downloads.

 

This is particularly dangerous on mobile phones, where users may install an application after seeing a convincing message on WhatsApp, Telegram, social media or email.

 

But do remember, a familiar logo does not prove that an application or website is genuine. Before downloading an AI application, go to the official app store or the service's official website rather than clicking an unsolicited link. Check the developer's name and look for signs that the service is legitimate. And never provide banking credentials or an OTP to activate an AI service.

 

Although Gen Z is particularly comfortable with AI, the privacy lesson applies to everyone. Children may use chatbots for homework. Teenagers may use them for relationship advice. Young professionals may turn to them for help with office documents or career decisions. Older users may ask AI to explain banking, health or government-related information.

 

The common risk is the same: the more human the conversation feels, the easier it is to forget that you are communicating with a technology service.

 

Parents should therefore discuss AI privacy with teenagers in the same way they discuss passwords, social media privacy and suspicious links.

 

The message should not be 'never use AI' or "Stay away from AI'. It should be 'know what should never be shared with any and every AI tool'.

 

AI is likely to become even more deeply embedded in everyday life. For India's young population in particular, it will increasingly be used for studying, work, communication, shopping and decision-making. That makes basic digital caution more important, not less.

 

The Kaspersky findings offer a useful warning: when people begin turning to AI for personal conversations, the line between using a tool and trusting a companion can become blurred.

 

There is nothing wrong with asking AI for ideas, explanations or help. But users should maintain a firm privacy boundary. “An interaction that feels personal should not be mistaken for a private or fully reliable conversation,” said Vladislav Tushkanov, group manager at Kaspersky AI Technology Research Centre. “Young users need to combine their confidence and curiosity with habits such as checking information, protecting personal data and thinking critically about the responses they receive.” 

 

Your chatbot may sound like a real Pookie, as you would have imagined. It may respond instantly and never interrupt you. It may even appear to understand your problem. But it is still a digital service. So, treat every piece of information you enter accordingly.

 

The safest habit is simple: share the minimum information needed to get the job done, remove identifying details wherever possible, verify important answers and never disclose secrets that could put your identity, money or the security of you and your human ‘Pookie’ at risk.