Before You Paste Into an AI Chatbot, Check the Privacy Terms

Prompts can contain personal data

A prompt may include names, customer records, unpublished work, medical details, or a screenshot with hidden account information. Once submitted, that material is processed under the service's terms and settings, which can differ between consumer, workplace, and paid accounts. Do not assume that closing a chat deletes every stored copy or that a prompt is automatically confidential.

Minimize before submitting

Ask whether the task can be completed with invented examples, a short excerpt, or de-identified information. Remove names, addresses, account numbers, and unrelated context. Review the service's current privacy controls for chat history, data retention, and use of submitted content, then confirm the account type your organization permits. A setting can change or apply only to future conversations, so read its explanation rather than relying on a label.

Keep accountability with people

If you use AI at work, follow data-classification rules and ask an administrator before sharing protected information. Do not upload confidential files simply because a tool offers a file button. Review generated answers for accuracy and disclosure risk before forwarding them. Privacy choices reduce unnecessary exposure, but they do not remove the need to check vendor terms, use approved tools, and maintain human oversight.

Sources and further reading

• NIST: AI Risk Management Framework
• FTC: How Websites and Apps Collect Information