Your Personal and Financial Details
This may seem obvious, but it’s the most critical rule. Never share personally identifiable information (PII) with a public chatbot. This includes your Aadhaar number, PAN, home address, personal phone number, or any login credentials. The same goes for financial
data like bank account numbers, credit card details, or salary information. While consumer AI apps can feel like private assistants, they are not bound by the same privacy laws as your bank or a healthcare provider. Every query you make can be logged and stored on company servers, creating a permanent record that could be exposed in a data breach.
Confidential Company and Client Information
Your company’s internal data is one of its most valuable assets. Pasting sensitive information into a chatbot is like shouting it in a public square. This includes unpublished financial reports, sales figures, marketing strategies, or client lists. Once that information is entered into a public AI model, the company loses control over it. It could be used to train future versions of the AI and, in a worst-case scenario, parts of that data could be inadvertently surfaced in a response to another user from a different company. This risks violating non-disclosure agreements and Indian laws like the Digital Personal Data Protection Act (DPDPA), which puts strict obligations on how companies handle data.
Proprietary Code and Internal Processes
For developers, engineers, and designers, using AI to debug code or brainstorm solutions is a huge productivity booster. However, pasting chunks of proprietary software code or detailed descriptions of your company's internal processes is a major risk. Many AI companies state that they can use your inputs to train their models. If your company’s unique code ends up in a training dataset, you could be jeopardizing its intellectual property (IP). Your organisation could lose its exclusive claim to that work, opening the door to commercial and legal challenges down the line. Always use an internal, sandboxed AI tool if your company provides one, or otherwise, keep proprietary details out of public platforms.
Internal Communications and Drafts
It can be tempting to paste a long email chain or a draft of a sensitive memo into a chatbot and ask it to summarize the key points. But these communications often contain a treasure trove of confidential context. Think about performance reviews, discussions about company strategy, or details of a pending negotiation. Even if names are removed, the context can reveal sensitive corporate information. Many companies now explicitly forbid using public AI for work-related information, and for good reason. Every interaction is recorded, and even deleted chats may not be permanently erased from server logs.
Personal Grievances and Opinions
Using a chatbot as a sounding board to vent about your boss, a coworker, or your job might feel harmless and private. It is not. These conversations can create a profile of you over time. Researchers have found that AI models can develop a 'conversational drift', subtly reflecting your biases or sentiments back to you in later chats. More importantly, these logs could potentially be accessed during a workplace investigation. What you share in what feels like a private moment could become evidence in an HR complaint or a performance review. It is always safer to discuss sensitive workplace issues with a trusted human colleague, a manager, or an HR representative.
