AI at Work: Think Before You Prompt Webinar

Generative AI is already part of the federal workday. Employees use it to draft memos, summarize reports, and brainstorm ideas, often without knowing where their prompts go or whether the tool is approved. This webinar gives you a plain-language guide to using AI at work without exposing government data, built for every role and requiring no cyber or technical background.

You'll see how a prompt travels from your keyboard to a vendor's servers and why unclassified does not mean safe to share. Real incidents from industry, government, and the courts show how ordinary work can turn into a data leak or a sanctioned filing. A “Would you paste this?” exercise and live polls will help you practice sorting safe prompts from risky ones.

We'll finish with a simple four-step prompt check (Classify, Check the tool, Clean the prompt, Confirm the output), habits for verifying AI output, and what to do if sensitive information has already been shared. You'll leave with a two-page handout and a script you can use today to find out what your agency allows.

Key Takeaways

• Understand how a prompt to a public AI tool can be stored, reviewed, or used for training and cannot be pulled back
• Identify the big four categories that should never go into an unapproved tool: PII, CUI, procurement-sensitive information, and pre-decisional information
• Distinguish agency-approved tools from “shadow AI” and know which office to ask when the answer is unclear
• Recognize that you are accountable for every word AI helps you produce, including errors, bias, and records obligations
• Know the four-step prompt check and the stop, don't delete, report, learn response if you slip