Cybersecurity Experts Flag Five Types of Sensitive Data to Keep Out of AI Chatbots

Generative AI chatbots are widely used to draft documents, provide customer service and organize data, but platforms may retain user inputs or use them to improve their systems. This creates risks involving personal data, trade secrets and regulatory compliance. The report did not identify the cybersecurity organization behind the warning, its publication date or any amount involved, but users should still treat chatbot interfaces as external systems.

Cybersecurity experts have most recently identified five categories of information that users should never enter: passwords, bank account and transaction details, customer data, legal documents and medical information. Users should limit prompts to the minimum data needed to complete a task. Companies should establish AI-use policies and introduce data masking, access controls and encryption to reduce the risk of leaks.